diff --git "a/A9AzT4oBgHgl3EQf__9t/content/tmp_files/load_file.txt" "b/A9AzT4oBgHgl3EQf__9t/content/tmp_files/load_file.txt" new file mode 100644--- /dev/null +++ "b/A9AzT4oBgHgl3EQf__9t/content/tmp_files/load_file.txt" @@ -0,0 +1,1616 @@ +filepath=/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf,len=1615 +page_content='High-level semantic feature matters few-shot unsupervised domain adaptation Lei Yu1, Wanqi Yang1*, Shengqi Huang1, Lei Wang2, Ming Yang1 1School of Computer and Electronic Information, Nanjing Normal University, China 2School of Computing and Information Technology, University of Wollongong, Australia yulei@njnu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='edu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='cn, yangwq@njnu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='edu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='cn, huangshengqi@njnu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='edu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='cn, leiw@uow.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='edu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='au, myang@njnu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='edu.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='cn Abstract In few-shot unsupervised domain adaptation (FS-UDA), most existing methods followed the few-shot learning (FSL) meth- ods to leverage the low-level local features (learned from con- ventional convolutional models, e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', ResNet) for classifica- tion.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' However, the goal of FS-UDA and FSL are relevant yet distinct, since FS-UDA aims to classify the samples in target domain rather than source domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We found that the local features are insufficient to FS-UDA, which could introduce noise or bias against classification, and not be used to effec- tively align the domains.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' To address the above issues, we aim to refine the local features to be more discriminative and rele- vant to classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Thus, we propose a novel task-specific semantic feature learning method (TSECS) for FS-UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' TSECS learns high-level semantic features for image-to-class similarity measurement.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Based on the high-level features, we design a cross-domain self-training strategy to leverage the few labeled samples in source domain to build the classi- fier in target domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In addition, we minimize the KL diver- gence of the high-level feature distributions between source and target domains to shorten the distance of the samples be- tween the two domains.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Extensive experiments on Domain- Net show that the proposed method significantly outperforms SOTA methods in FS-UDA by a large margin (i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', ∼ 10%).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' keywords Few-shot unsupervised domain adaptation, image-to-class similarity, high-level semantic features, cross-domain self- training, cross-attention.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Introduction Currently, a setting namely few-shot unsupervised domain adaptation (FS-UDA) (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021)(Yang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2022), which utilizes few labeled data in source domain to train a model to classify unlabeled data in target domain, owns its potential feasibility.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Typically, a FS-UDA model could learn general knowledge from base classes during training to guide classification in novel classes during testing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' It is known that both insufficient labels in source domain and large domain shift make FS-UDA as a challenging task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Previous studies, e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', IMSE (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021), first fol- lowed several few-shot learning (FSL) methods (Li et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The corresponding author is Wanqi Yang.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Copyright © 2023, Association for the Advancement of Artificial Intelligence (www.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='aaai.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='org).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' All rights reserved.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Figure 1: A 5-way 1-shot task for FS-UDA where the sup- port set includes five classes and one sample for each class.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The figure shows the similarity of query images to every support classes and the spatial similarity of query images to the predicted support class.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We found using local fea- tures could cause some inaccurate regions of query images to match the incorrect classes, while our semantic features make the object region in query images similar with their true class, thus achieving correct classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019)(Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017) to learn the local features by us- ing convolutional models (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', ResNet) and then leveraged them to learn image-to-class similarity pattern for classifica- tion.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' However, we wish to clarify that the goal of FS-UDA and FSL are relevant yet distinct, since both of them suf- fer from insufficient labeled training data whereas FS-UDA aims to classify the samples in target domain rather than source domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' As shown in Fig.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 1, by visualizing the spatial similarity of query images to predicted support classes, we found using local features causes the inaccurate regions of query images to match incorrect classes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' This reason might be that few labeled samples and large domain shift between the support and query sets simultaneously result in the con- ventional local features in FSL to fail in classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In this sense, the local features are insufficient to FS-UDA, which could introduce noise or bias against the classification in tar- get domain and not be used to effectively align the domains.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' To address this issue, we aim to refine the low-level local arXiv:2301.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='01956v1 [cs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='CV] 5 Jan 2023 support set in the source domain (sketch) sailboat bed glasses television snowman query set in the target domain (clipart) local features semantic features (ours) 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='6 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='6 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='5 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='5 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='4 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='4 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='3 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='3 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='2 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='2 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='1 bed local features semantic features (ours) 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='6 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='6 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='5 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='5 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='4 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='4 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='3 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='3 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='2 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='2 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='1 television as saFigure 2: Illustration of the process for cross-domain self- training in TSECS.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Different shapes represent different do- mains.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We first select the ‘confidence’ target samples (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', a) that are very similar to support classes, and then regard them as the new class prototypes to further classify the other target samples (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', b, c).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' This process is executed itera- tively with using class matching loss to narrow the distance of query images and their most similar support classes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' features to be more discriminative and relevant to classifica- tion, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', high-level semantic features, and meanwhile align the semantic features for domain adaptation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Therefore, we propose a novel task-specific semantic feature method (TSECS) that learns the semantic features for each task by clustering the local features of support set and query set.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' To obtain the related semantics from previous tasks, the cluster centroids of the current task are then fused by cross-attention with that of the previous task to generate high-level semantic features to boost classification performance.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moreover, for the domain shift between source and tar- get domains, many domain adaptation methods (Saito et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2018)(Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017)(Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2014) reduced the dis- tribution discrepancy between domains by using a discrim- inator to adverse against feature embedding.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' However, this way could fail in aligning the samples of the same class be- tween domains due to label missing in target domain, which could make the classes of two domains mismatched and thus affect the classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Therefore, we aim to align the high- level semantic features by minimizing the KL divergence of the semantic feature distributions between domains, and meanwhile design a cross-domain self-training strategy to train the classifier in target domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We hypothesis that there are usually several ‘confidence’ samples in target domain that could be classified correctly by support set in source domain, in other words, they are very similar to their class prototypes in source domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Mean- while, the target domain samples in the same class are more similar to each other than that of other classes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Based on this, we regard these ‘confidence’ samples in the target domain as new prototypes of the classes, which replace those from the support set of source domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' As shown in Fig.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2, several ‘confidence’ samples (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', a) can be selected as prototypes of their similar classes for classification (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', b and c) in tar- get domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moreover, the process is conducted iteratively by using class matching loss for better domain alignment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In sum, we propose the novel method, namely TSECS, for FS-UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' It refines the local features of convolutional network to generate specific semantic features of each task, and meanwhile perform cross-domain self-training to trans- port labels from support set in the source domain to query set in the target domain to effectively classify the samples in target domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Our contributions can be summarized as: (1) A novel solution for FS-UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' TSECS aims to learn high-level semantic features for classification and do- main alignment, which could be regarded as a more ef- fective and efficient way than using local features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (2) Task-specific semantic embedding for few-shot set- ting.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' It can be seamlessly add to existing FSL/FS-UDA models, which could alleviate the bias of classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (3) Cross-domain self-training for domain alignment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' It is designed to bring the samples of the same class close, which could guide effective domain alignment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We conduct extensive experiments on DomainNet.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Our method significantly outperforms SOTA methods in FS- UDA by a large margin up to ∼ 10%.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Related Works Unsupervised domain adaptation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The conventional UDA methods aim to reduce discrepancy between source domain and target domain in the feature space and utilize suffi- ciently labeled source domain data to classify data from tar- get domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The difference between unsupervised domain adaptation methods often lies in the evaluation of domain discrepancy and the objective function of model training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Several researchers (Long et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2015)(Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2014) minimize the feature discrepancy by using maximum mean discrepancy to measure the discrepancy between the distri- bution of domains.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moreover, adversarial training (Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017)(Ganin et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2016) to learn domain-invariant fea- tures is usually used to tackle domain shift.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Several meth- ods (Tang, Chen, and Jia 2020)(Zou et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019)(Zou et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2018)(Kim et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021)train the classifier in both source do- main and target domain and utilize pseudo-labels from target domain to calculate classification loss.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Overall, these UDA methods all require sufficiently labeled source domain data to realize domain alignment and classification, but they per- form poor when labeled source domain data are insufficient.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Few-shot learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Few-shot learning has two main streams, metric-based and optimization-based approaches.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Optimization-based methods (Bertinetto et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019)(Finn, Abbeel, and Levine 2017)(Ravi and Larochelle 2017) usu- ally train a meta learner over auxiliary dataset to learn a general initialization model, which can fine-tune and adapt to new tasks very soon.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The main purpose of metric- based methods (Li et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019)(Snell, Swersky, and Zemel 2017)(Vinyals et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2016)(Ye et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020) is that learn a gen- eralizable feature embedding for metric learning, which can immediately adapt to new tasks without any fine-tune and retraining.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Typically, ProtoNet (Snell, Swersky, and Zemel 2017) learns the class prototypes in the support set and clas- sifies the query images based on the maximum similarity to these prototypes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Other than these metric-based methods on feature maps, many methods on local features have ap- peared.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' DN4 (Li et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019) utilizes large amount of local features to measure the similarity between support and query select \'confidence" sanples use new prototypes for as new prototypes classification in target domain O b O 0 00 00 lass natching loss 0 00 dims prototypes doeifiad query imega [sonmce dm ngin) (trt domain) at din) (trt domain)sets instead of flattening the feature map into a long vec- tor.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Based on local features, DeepEMD (Zhang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020) adopts Earth Mover’s Distance distance to measure the re- lationship between query and support sets.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Furthermore, a few recent works focus on the issue of cross-domain FSL in which domain shift exists between data of meta tasks and new tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The baseline models (Chen et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019) are used to do cross-domain FSL.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' LFT (Tseng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020) performs adaptive feature transformation to tackle the domain shift.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Few-shot unsupervised domain adaptation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Compared with UDA, FS-UDA is to deal with many UDA tasks by leveraging few labeled source domain samples for each.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' And compared with cross-domain FSL, FS-UDA are capable of handling the circumstances of no available labels in the tar- get domain, and large domain gap between the support and query sets in every task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For the one-shot UDA (Luo et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020), it deals with the case that only one unlabeled target sample is available, but does not require the source domain to be few-shot, which is different from ours.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Recently, there are a few attempts in FS-UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' PCS (Yue et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) per- forms prototype self-supervised learning in cross-domain, but they require enough unlabeled source samples to learn prototypes and ignore task-level transfer, which is also dif- ferent from ours.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' meta-FUDA (Yang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2022) lever- ages meta learning-based optimization to perform task-level transfer and domain-level transfer jointly.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' IMSE (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) utilizes local features to learn similarity patterns for cross-domain similarity measurement.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' However, they did not consider that local features could bring the noise or bias to affect classification and domain alignment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Thus, we pro- pose task-specific semantic features to solve this problem.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Methodology Problem Definition A N-way, K-shot FS-UDA task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Table 1 shows the main symbols used in this paper.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The FS-UDA setting includes two domains: a source domain S and a target domain T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' A N-way, K-shot FS-UDA task includes a support set XS from S and a query set QT from T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The support set XS contains N classes and K samples per class in the source domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The query set QT contains the same N classes as in XS and Nq target domain samples per class.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' To classify query images in QT to the correct class in XS, it is popular to train a general model from base classes to adapt to handle new N-way, K-shot FS-UDA tasks for testing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Auxiliary dataset and episodic training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' As in (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021), the base classes are collected from an auxil- iary dataset Daux to perform episodic training to learn the general model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Note that the base classes in Daux are com- pletely different from new classes in testing tasks, which are unseen during episodic training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moreover, Daux includes labeled source domain data and unlabeled target domain data for FS-UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We construct large amounts of episodes, each containing {XS, QS, QT } as in (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021), to simulate the testing tasks for task-level generalization.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Note that QS is introduced into episodic training to calculate clas- sification loss and perform domain alignment with QT .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The flowchart of our method.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Fig.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 3 illustrates our Table 1: Notations Notations Descriptions N ∈ R The number of classes in the task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' K ∈ R The number of samples per class in support set.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' XS, QS, QT Support set of source domain, and query sets of source domain and target domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' H, W, d ∈ R The height, width, and channel of feature map.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' L ∈ RHW ×d The local feature vectors in the feature map.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' k ∈ R The number of semantic clusters for an episode.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' C ∈ Rk×d The centroids of the clusters.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' F, ˆF, The semantic feature map, semantic features and ˆFXS, ˆFQS, ˆFQT the parts of support and query sets in both domains.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' M c q ∈ RH×W ×N The 3-D similarity matrix for classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' pc q ∈ RKHW Similarity pattern vectors of a query image q pi q ∈ RHW with a support class c and a support image i, ppos q , pneg q ∈ RKHW and the most similar class and the second one for q.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' µA, µB ∈ RHW ×d The mean of semantic features or similarity patterns.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ΣA, ΣB ∈ RHW ×HW Covariance matrix of semantic features or similarity patterns.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' λsfa, λspa, λclm Weight parameters of three loss terms in Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (6).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' method for 5-way, 1-shot FS-UDA tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In each episode, a support set (XS) and two query sets (QS and QT ) are first through the convolution network (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', ResNet) to extract their local features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Then, the task-specific semantic embed- ding module refines the local features to generate semantic features, which is computational efficient due to dimension reduction.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Also, based on semantic features of QS and QT , we leverage their similarity patterns (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) to calculate image-to-class similarity for classification with the loss Lcls.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' To improve its performance, cross-domain self- training module is performed to introduce the class proto- types of target domain and train a target domain classifier with a class matching loss Lclm.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In addition, the seman- tic features and similarity patterns from both domains are further aligned by calculating their alignment losses Lsfa and Lspa, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Finally, the losses above are back- propagated to update our model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' After episodic training over all episodes, we utilize the learned model to test new FS- UDA tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Then, we calculate the averaged classification accuracy on these tasks for performance evaluation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Task-specific Semantic Feature Learning Most FSL methods and FS-UDA methods learned local fea- tures from convolutional networks for classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' How- ever, we found that the local features could introduce noise or bias that is valid for classification and domain alignment.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Thus, we aim to refine the local features to generate high- level semantic features for each task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In the following, we will introduce our semantic feature embedding module.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' First of all, in each episode, all local features L ∈ R(|XS|+|QS|+|QT |)HW ×d are extracted from the convolu- tional network, where | · | is the number of samples in a set.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Then, we cluster the local features to generate different semantic clusters for support set and query set, respectively, since clustering the two sets together could result in the clus- ters that relate to the domains due to the presence of large do- main gap.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For simplification, we adopt K-means for cluster- ing, and meanwhile utilize the singular value decomposition (SVD) to adaptively take the number of eigenvalues greater than a certain threshold as the cluster number k (k ≪ d) for each task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Afterwards, we calculate the task-specific seman- Figure 3: Illustration of our method training per episode for 1-shot FS-UDA tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' First, support classes and query images from both domains are through a convolution network to extract their local features, followed by the task-specific semantic embedding module to learn high-level semantic features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Then, these semantic features are fed into the cross-domain self- training module to update the class prototypes for target domain classification and calculate the class matching loss Lclm.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Meanwhile, these semantic features are also used to generate similarity patterns in IMSE (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) for classification loss Lcls.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In addition, both semantic features and similarity patterns from both domains are aligned by the domain alignment module with the alignment losses Lsfa and Lspa, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Finally, all the losses are backpropagated to update our model.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' tic feature map F ∈ R(|XS|+|QS|+|QT |)HW ×k by measuring the Cosine similarity between the local features L and the centroids C ∈ Rk×d of all semantic clusters, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', F = L ||L||2 · C⊤ ||C||2 .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Finally, we split F to 2×2 blocks based on height and weight dimension of the feature map, and then concatenate the four blocks together along the channel to generate se- mantic features ˆF ∈ R 1 4 (|XS|+|QS|+|QT |)HW ×4k.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' This is a simple yet effective way to maintain discriminative ability and spatial information of semantic features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moreover, to leverage the semantics from previous tasks to guide the semantic feature learning of the current task, we utilize the centroids of previous clusters to update the initial- ization of clustering centroids by cross-attention (Li et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' This makes K-means clustering converge rapidly.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' After obtaining the semantic features ˆF, we use them for domain alignment and classification.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Firstly, ˆF is partitioned into ˆFXS, ˆFQS, ˆFQT along with the first dimension.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Then, we align ˆFQS and ˆFQT by minimizing the KL divergence of their distributions that will be introduced later.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Meanwhile, we utilize ˆFXS, ˆFQS and ˆFQT to build 3-D similarity matrix M c q (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) between support and query sets.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Fi- nally, we calculate the similarity pattern pc q (measuring the similarity between query sample q and support class c) for classification (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The classification loss us- ing cross-entropy can be written by: Lcls = − 1 |QS| � q∈QS log( exp(1 · pc q) �K i=1 exp(1 · piq) ) (1) Cross-domain Self-training Since there is large domain shift between source and target domains, as well as label missing in target domain, adver- sarial domain adaptation on low-level local features cannot make samples of the same class between domains close, and thus could make the classes of two domains mismatched.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' To alleviate the mismatching issue, we aim to find the most similar ‘confidence’ samples in QT with XS to guide classification in target domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We assume that it usually exists that the ‘confidence’ samples in QT could be clas- sified correctly by XS, when the distributions between do- mains are aligned.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We iteratively select the ‘confidence’ samples in QT as the new prototypes to replace that in XS for classification, as shown in Fig.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We call the process as cross-domain self-training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The process can find more ‘con- fidence’ samples from QT than that in XS for the same class, which could correct some misclassified samples in QT , thereby lightening the impact of domain gap.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moreover, to improve the performance of the target do- main classifier, we aim to make target domain samples q in QT closer to their most similar class and meanwhile far away from the other classes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Thus, we first calculate its sim- ilarity patterns ppos q (with the most similar class) and pneg q (with the second similar class), and then design the class matching loss with a margin m, which can be written by Lclm = � q∈QT max(softmax(pneg q )−softmax(ppos q )+m, 0), (2) where the similarity to the most similar class should be greater by m than the second similar class.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Two-level Domain Alignment Conventional adversarial domain adaptation methods (Ganin et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2016)(Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017) iteratively train a discriminator to align the distribution of domains by adver- sarial training among tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' However,' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' they cannot be used to align the semantic features,' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' because our semantic features are relevant to tasks,' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' the semantics of the same channel ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Task-specific semantic embedding ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Local features ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Semantic feature maps ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='High-level semantic features ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Support class ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='(Source domain) ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Similarity patterns in IMSE ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Qurey image ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='MH ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='(Target domain) ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='conv ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Lcls ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Classification loss ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Query image ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='(Source domain) ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Split into 2 x 2 blocks ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='I Update the class ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='and concatenate them ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='prototypes ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='k ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Cross-domain self-training ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Domain alignment ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='I Semantic features ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content="Similarity paterns' I " metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='Centers of k clusters ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='class prototype ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='( confidence ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='sample ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='1 ' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='KL(*，' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='*) KL(*，' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='*) Cp Source domain in support set Clustering Target domain in query set Lclm Lsfa Lspa Class matching loss + Source domain in query set Aligment loss Loss backpropagationcould be varied for different tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Meanwhile, symmetrical alignment could bring the inference information of the target domain to the source domain (Li et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Thus, we use asymmetrical KL divergence to align the distribution of domains on both semantic features and similarity patterns within a task.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Then, KL divergence can be calculated by: KL(A, B) =1 2 � tr(Σ-1 AΣB) + ln(ΣA ΣB ) +(µA − µB)Σ-1 A(µA − µB)⊤ − d � , (3) where µA, µB, ΣA and ΣB are the mean vectors and the co- variance matrices of sample matrix A and B, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Thus, we minimize the KL divergence between semantic features ˆHQS and ˆHQT by Lsfa = KL( ˆFQS, ˆFQT ).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (4) Meanwhile, we also minimize the KL divergence to align the similarity patterns {pc qS} of QS and {pc qT } of QT with class c, which can be written by Lspa = N � c=1 KL({pc qS}, {pc qT }).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (5) In sum, we combine all the above losses, w.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='r.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='t.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' classifi- cation (Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (1)), class matching (Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (2)) and KL-based do- main alignment (Eqs.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (4) and (5)) to train our model on many episodes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The total objective function can be written by: min Lcls + λsfaLsfa + λspaLspa + λclmLclm, (6) where the hyper-parameters λsfa, λspa and λclm are intro- duced to balance the effect of different loss terms.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Experiment DomainNet dataset.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We conduct extensive experiments on a multi-domain benchmark dataset DomainNet to demonstrate the efficacy of our method.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' It was released in 2019 for the re- search of multi-source domain adaptation (Peng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' It contains 345 categories and six domains per category, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', quickdraw, clipart, real, sketch, painting and infograph do- mains.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In our experiments, we follow the setting of IMSE in (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) to remove data insufficient domain infograph.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' There are 20 combinations totally for evaluation, and the dataset is split into 217, 43 and 48 categories for episodic training, model validation and testing new tasks, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Note that in each split every category contains the five-domain images.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Network architecture and setting.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We employ ResNet- 12 as the backbone of feature embedding network, which is widely used in few-shot learning (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) (Gi- daris et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We obtain semantic features by first clus- tering the local features from each class of support set and two query sets and then concatenating them.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' During this pro- cess, we adopt cross-attention that consists of three convo- lution parameters to generate (Q, K, V ) for attention cal- culation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In cross-domain self-training module, we set the threshold 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='7 of similarity score to select the ‘confidence’ samples in target domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The margin m in Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (2) is empir- ically set to 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In addition, we follow the setting of IMSE (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) to obtain similarity patterns.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The hyper- parameters λsfa, λspa and λclm are set to 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='1, 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='05 and 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='01, by grid search, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Model training, validation and testing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' To improve the performance, before episodic training, the feature embed- ding network is pretrained by using source domain data in the auxiliary dataset, as in (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Afterwards, we perform episodic training on 280 episodes, following the setting of (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' During episode training, the total loss in Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (6) is minimized to optimize the network parameters for each episode.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Also, we employ Adam opti- mizer with an initial learning rate of 10-4, and meanwhile re- duce the learning rate by half every 280 episodes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For model validation, we compare the performance of different model parameters on 100 tasks, which is randomly sampled from the validate set containing 43 categories.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Then, we select the model parameters with the best validation accuracy for test- ing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' During the testing, we randomly select 3000 tasks to calculate the averaged top-1 accuracy on these tasks as the evaluation criterion.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Comparison Experiments for FS-UDA We conduct extensive experiments on DomainNet to com- pare our method with five FSL methods (ProtoNet (Snell, Swersky, and Zemel 2017), DN4 (Li et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019), ADM (Li et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020), FEAT (Ye et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020), DeepEMD (Zhang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020)), three UDA methods, (MCD (Saito et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2018), ADDA (Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017), DWT (Roy et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019)), their combinations and the most related method IMSE (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For fair comparison, the results of these above methods are all reported from (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) with the same setting.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moreover, we also modify IMSE by using our semantic features for classification and domain adver- sary, namely IMSE+TSE.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For fair comparison, these com- pared methods also pretrain the embedding network before episodic training, and they are trained on 1000 episodes.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Comparison analysis.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Table 2 shows the results of all the compared methods for 20 cross-domain combinations, which records the averaged classification accuracy of tar- get domain samples over 3000 5-way 1-shot/5-shot FS- UDA tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' As observed, our TSECS achieves the best per- formance for all combinations and their average.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Specifi- cally, the UDA and FSL baselines in the first two parts per- form the worst.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In the third part, the combination methods with ADDA (Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017) perform domain adversarial training each episode, thus generally better than the above two parts, but still inferior to IMSE (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021) and our TSECS.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' This is because the combination methods only perform domain alignment based on original feature maps, not considering the alignment of similarity patterns (related to classification predictions).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Also, IMSE is worse than IMSE+TSE, which indicates high-level semantic fea- tures are more effective for FS-UDA than local features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' However, they are still much worse than our method, show- ing the efficacy of high-level semantic features and cross- domain self-training for FS-UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' On the other hand, we can see that the 20 cross-domain combinations have considerably different performances.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' This is because several domains (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', quickdraw) are sig- nificantly different from other domains, while several other domains (e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='g.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' real, clipart) are with the similar styles and features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Thus, for most compared methods, the perfor- Table 2: Comparison of our method with the related methods for 5-way 1-shot or 5-shot FS-UDA tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The first three blocks and IMSE are reported from (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021), while the last two are the variant of IMSE we designed and ours, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Each row represents the accuracy (%) of a compared method adapting between two domains, where the skt, rel, qdr, pnt, and cli denote the sketch, real, quickdraw, painting, and clipart domains in DomainNet, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The best results are in bold.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 5-way, 1-shot Methods skt ←→ rel skt ←→ qdr skt ←→ pnt skt ←→ cli rel ←→ qdr rel ←→ pnt rel ←→ cli qdr ←→ pnt qdr ←→ cli pnt ←→ cli avg → / ← → / ← → / ← → / ← → / ← → / ← → / ← → / ← → / ← → / ← MCD 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='07/37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74 38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='90/34.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51 39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='31/35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43/38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98 24.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='17/29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='85 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='36/47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='32 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='71/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='68 26.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14/25.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='00/34.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='69 39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='49/37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28 38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21 ADDA 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06 38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='42/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='88 50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='67/47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='16 31.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='78/35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='47 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30/47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='66 26.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='57/27.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51/32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='19 39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76/41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91 DWT 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43/38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='67 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='94/38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='00 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='73/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02/50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='69 29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82/29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='99 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='81/50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='55 24.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33/25.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='90 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='47/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='55/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='38 ProtoNet 50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='48/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='20/32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='63 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='69 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='17 32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='48/25.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06/50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='95 22.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='55/28.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76 36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13/41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21 DN4 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='42/47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='29 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46/35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='55 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25 33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='41/27.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='48 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='90/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84 22.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82/29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='11 36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='88/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='61 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='42/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='81 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='61 ADM 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='36/42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='27 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45/30.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='62/36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='34/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64 32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77/24.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='37 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='8/50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15 21.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43/30.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='12 35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='49/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='11 FEAT 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='72/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='66 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='29/35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='09/42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='99 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='69/50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59 33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='81/27.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='58 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='31 23.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10/29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39 37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='27/42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='54 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15/44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='49 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14 DeepEMD 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='12/34.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='89 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10/49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56 34.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28/28.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='73/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='26 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91 22.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86/28.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='79 37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='65/42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='92 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='11/44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='38 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46 ADDA+ProtoNet 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='79/35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02/41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='68/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91 37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28/27.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='68 50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='04/49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='68 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='58 23.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='72/32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03 38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='54/44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06/41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 ADDA+DN4 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='04/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64/36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='38/47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='97/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28 34.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='80/29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='09/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='05 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='81/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08 23.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='67/31.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='62 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25/44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='65 ADDA+ADM 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='87/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91/32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='38 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='48/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='37 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='81/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14 35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86/28.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='88/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='61 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='95/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='29 23.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='95/33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76 ADDA+FEAT 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='72/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='00/36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='94 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='01 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10 36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='32/30.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='50 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='36 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86 24.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76/35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='38 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='66/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='92 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='20 ADDA+DeepEMD 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98/47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='55 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64/36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='19 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='29/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93/50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='47/30.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='32 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='80 23.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46/32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='89 39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39/44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='65 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='75 IMSE 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='71/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91 50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='36/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='35 59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='44/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43/36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='55 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='09/57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98 30.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='73/38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='70 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='94/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='47 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='42/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86 IMSE+TSE 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='71/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='78/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='57 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='50/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='48/49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='44/57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60/59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52 37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='94/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='22 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='19/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='01 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='79 TSECS (ours) 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='00/58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='22 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='97 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='70 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45/64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='66/49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76/63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='18 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98/67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='89 38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='26/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51/69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='20 5-way, 5-shot Methods skt ←→ rel skt ←→ qdr skt ←→ pnt skt ←→ cli rel ←→ qdr rel ←→ pnt rel ←→ cli qdr ←→ pnt qdr ←→ cli pnt ←→ cli avg → / ← → / ← → / ← → / ← → / ← → / ← → / ← → / ← → / ← → / ← MCD 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='42/47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='73 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='73 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='63/47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='75 72.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='17/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='23 28.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02/33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74/66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='80/63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='07 28.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='71/29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='17 50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46/45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='99/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='65 ADDA 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='66 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='37/42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='61/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='95 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='57/65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='81 35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='94/36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='87 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='11/63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56 59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='16/65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='7 723.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='16/33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='50 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='94/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76 DWT 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='75/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='85 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40/50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='87/59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33 36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='19/36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='26/68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='72 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='92/67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28 22.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64/32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='34 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='88/50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='47 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74 ProtoNet 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='07/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='65/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='75 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43/62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='62 37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77/31.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='01 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='73/66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='85 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52/66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 20.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74/30.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='55 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='49/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='92 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='80 DN4 (Li et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019) 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='89/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='96 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='23/38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='68 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='57/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='62 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='88/58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33 37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25/29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03/64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='72 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10/62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25 23.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86/33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77/49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46 50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='63/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='41 ADM 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='20 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15/35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='69 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60 71.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='73/63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='42 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='61/24.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83 59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='48/69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='17 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='54/67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39 21.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13/38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74/58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='36 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='34/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='78 FEAT 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91/58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='27/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='97 59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='01/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='44 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='37/65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='95 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='71/28.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='65 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='85/71.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76/68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='96 23.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='73/34.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='95/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='78 DeepEMD 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='96/58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='11 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='34/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='70 59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='31/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60 70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56/64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60 39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='70/29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='95 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='99/70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='07/69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06 23.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86/34.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='34 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='48/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='61 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93 ADDA+ProtoNet 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='11/58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='72 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='92/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='23/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='90 68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='44/61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='59/38.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='94/69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='47 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30/66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10 25.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45/41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='67/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='22 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='20/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='65 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52 ADDA+DN4 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='37/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='12 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51/49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='69 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93/58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39 36.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='92/31.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08/65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='92 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74/63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13 25.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='36/34.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='23 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='19 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='16/49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='62 50.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33 ADDA+ADM 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='65 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56/33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='42 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='85 70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='70/63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='57 39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93/27.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='17 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='63/68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='70 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='96/67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='29 21.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91/39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='12 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='96/59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='57/53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='27 ADDA+FEAT 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='80/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='71 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33/43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='34 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='32/58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08 70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='06/64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='57 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13/35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='62 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='09/70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='32 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46/67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77 29.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08/44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='62/63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='38 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='34/52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='56 ADDA+DeepEMD 68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52/59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='78/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='18/57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86 70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83/65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='63/32.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='18 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82/71.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='54 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51/69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21 26.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='89/42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33 47.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='00/57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='92 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='81/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='23 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='49 IMSE 70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46/61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='09 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='57/46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='30/59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15 76.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13/67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='27 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='07/40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='17 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='41/70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='63 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60/71.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76 33.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='44/48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='89 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='38/65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='90 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28/56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74 59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='60 IMSE+TSE 72.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='75/62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='24 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='49/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='04 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86/61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='10 77.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39/69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='87 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='88/54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='48 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='97/72.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='86/72.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='49 37.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43/51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='66 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='43/67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40/57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='89 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='76 TSECS (ours) 78.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='23/70.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='44 77.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='90/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='70/68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='03 83.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82/74.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33/55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='16 68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40/79.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74 78.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='23/77.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='69 39.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74/63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='02 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='99/80.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='31 73.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='67/61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='63 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25 Table 3: Ablation study (%) of the modules designed in TSECS, where the FS-UDA tasks are evaluated from a do- main (sketch) to the other four domains in DomainNet.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Components Target Domains TSE catt CS cli rel qdr pnt ✓ 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='00 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='21 51.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='62 ✓ 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='07 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='31 41.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='93 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='66 ✓ ✓ 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='74 60.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='54 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='23 ✓ ✓ 68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25 61.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='15 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='31 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='34 ✓ ✓ ✓ 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='00 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51 mance becomes relatively low when the domain gap is large.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For example, from quickdraw to painting, it performs the worst in all the other combinations because of larger domain gap, but our TSECS outperforms IMSE and the other com- pared methods by 8% and 12%, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We found that our method has the larger performance improvement over IMSE, for these combinations containing quickdraw, which shows the efficacy of our method for large domain gap.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Also, like TSECS, IMSE+TSE performs much better than IMSE for large domain gap, which indicates the high-level seman- tic features could conduct domain adaptation better than lo- cal features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In sum, these results reflect the advantages of our TSECS to deal with domain shift and task generaliza- tion in FS-UDA, no matter how large the domain gap is.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ablation study of our method.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We conduct various ex- periments on DomainNet to evaluate the effect of our mod- ules: task-specific semantic embedding (TSE), cross-domain self-training (CS) and cross-attention in TSE (catt).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The ac- curacies on the four target domains are reported in Table 3.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' As seen, our method achieve the best performance when three modules are all used.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The performance of the single CS is the worst that shows that local features cannot align the distributions of the two domains, thus affecting cross- domain self-training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The module TSE is introduced into four combinations, all improving the performance, which validates the efficacy of our task-specific semantic features for FS-UDA again.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Also, the addition of cross-attention into TSE will further improve the performance, which can help discover more semantics from previous tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ablation study of different losses.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We conduct various experiments on DomainNet to further evaluate the effect of different losses in Eq.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' (6).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Besides the classification loss (Lcls), we combine the remaining three loss terms: 1) se- mantic features alignment loss (Lsfa), 2) similarity pattern alignment loss (Lspa), and 3) class matching loss (Lclm).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We evaluate 5-way 1-shot FS-UDA tasks from sketch to the other four domains, respectively, and their accuracies are re- ported in Table 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' As observed, the more the number of loss terms involved, the higher the accuracy.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The combination of all the three losses is the best.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For the single loss, both Lsfa Table 4: Ablation study (%) of the three losses designed in TSECS, where the FS-UDA tasks are evaluated from a do- main (sketch) to the other four domains in DomainNet.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Components Target Domains Lsfa Lspa Lclm cli rel qdr pnt ✓ 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='67 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='84 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91 43.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28 ✓ 64.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='28 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='32 52.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='11 42.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='46 ✓ 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83 58.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='29 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25 ✓ ✓ 66.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='64 57.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='41 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='40 ✓ ✓ 68.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='04 63.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='98 59.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='13 55.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='39 ✓ ✓ 67.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='61 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='47 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='07 54.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14 ✓ ✓ ✓ 69.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='45 65.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='00 62.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='25 56.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='51 Figure 4: Comparison of introducing our TSE module or not into two FSL methods with ADDA (Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017) com- bined, i.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='e.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', ADDA+ProtoNet and ADDA+DN4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Lclm perform better than Lspa, and their combination is also considerably better than the other paired combinations, showing the efficacy of semantic feature domain alignment and class matching in target domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Based on the above, adding Lspa further improves the performance, indicating positive effect of aligning the similarity patterns.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Evaluation on the effect of our task-specific se- mantic embedding module on two FSL methods with ADDA (Tzeng et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017) combined.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Compared with ADDA+DN4 and ADDA+ProtoNet, we add our semantic embedding module (TSE) with the loss Lsfa into their fea- ture embedding models, and test them on 3000 new 5-way 1/5-shot FS-UDA tasks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For simplification and clarification, we calculate the averaged accuracies from every domain to the other four domains and show them in Fig.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 4.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' As seen, the methods using TSE generally perform better than that without it, which validates that the semantic embedding in TSE could generate more discriminative semantic features for classification than original local features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In addition, the performances of these methods are still far from our method because using ADDA is insufficient to align the domains and could result in class mismatching, but our method can effec- tively solve it by cross-domain self-training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Evaluation of dataset generalization.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We evaluate the generalization of our model trained on DomainNet to adapt to a substantially different dataset miniImageNet.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We mod- ify miniImageNet by transferring a half of real images (rel) into sketch images (skt) by MUNIT (Huang et al.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2018) to Table 5: Evaluation (%) of dataset generalization for 5-way 1-shot FS-UDA tasks between domains real and sketch, per- forming episodic training on DomainNet and testing on ex- panded dataset miniImageNet.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Methods skt → rel rel → skt ADDA+DN4 44.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='01 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='87 40.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='61 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='90 ADDA+DeepEMD 46.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='14 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='82 45.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='91 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='77 IMSE 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='78 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='78 48.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='52 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='81 TSECS (ours) 53.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='33 ± 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='08 49.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='83 ± 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='96 Figure 5: The tSNE visualization of our TSECS using cross- domain self-training or not for a 5-way 5-shot FS-UDA task from sketch to clipart.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The samples with different colors be- long to different classes, and the stars in the left and right figures represent the class centroids of support set and se- lected target domain query samples, respectively.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' produce two domains for FS-UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We compare our method with ADDA+DN4, ADDA+DeepEMD and IMSE for 5-way 1-shot FS-UDA tasks for rel ↔ skt.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' The results are shown as Table 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' As observed, our method outperforms other methods, specially for ske → rel.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' For rel → skt, our method is slightly better than IMSE, because the style of sketch im- ages in miniImageNet is relatively different from that in Do- mainNet, which could effect the learned semantic features.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Visualization of our method using cross-domain self- training or not.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We illustrate the tSNE results of a 5-way 5- shot FS-UDA task from sketch to clipart in Fig.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 5.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Note that the class prototypes in the left subfigure belong to the sup- port set in source domain, while those in the right subfigure are generated by ‘confidence’ samples in target domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' It is obvious that two class prototypes in the left subfigure are fully overlapped so that many samples could not be correctly classified.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In contrast, the right subfigure has the better class prototypes, and samples from different classes are more dis- tinguishable.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' This shows the efficacy of our cross-domain self-training that finds ‘confidence’ samples to train the tar- get domain classifier and uses class matching loss Lclm to shorten the distance of samples of the same class.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Conclusion In this paper, we propose a novel method TSECS for FS- UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' We extract high-level semantic features than local fea- tures to measure the similarity of query images in target do- main to support classes in source domain.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moreover, we de- sign cross-domain self-training to train a target domain clas- sifier.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In addition, asymmetrical KL-divergence is used to align the semantic features between domains.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Extensive ex- periments on DomainNet show the efficacy of our TSECS, significantly improving the performance for FS-UDA.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ProtoNet/1-shot .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='- DN4/1-shot 701 701 60 60 50 50 40 40 30 30 skt cli rel qdr pnt skt clirel qdr pnt ProtoNet/5-shot DN4/5-shot 70 70 60 60 50 50 40 40 30 30 skt clirel Ipb pnt skt clirel qdr pnt ADDA+ProtoNet ADDA+DN4 ADDA+TSE+ProtoNet ADDA+TSE+DN4 ours oursTSECS (no CS) TSECS 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='0 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='0 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='8 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='8 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='6 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='6 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='4 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='4 .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ★ 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='2 .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='2 .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ： .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='0 .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='0 .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' .' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='0 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='2 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='4 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='6 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='8 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='0 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='0 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='2 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='4 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='6 0.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='8 1.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='0Acknowledgments Wanqi Yang and Ming Yang are supported by Na- tional Natural Science Foundation of China (Grant Nos.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 62076135, 62276138, 61876087).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Lei Wang is supported by an Australian Research Council Discovery Project (No.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' DP200101289) funded by the Australian Government.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' References Bertinetto, L.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Henriques, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' F.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Torr, P.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Vedaldi, A.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Meta-learning with differentiable closed-form solvers.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In International Conference on Learning Represen- tations, 1–8.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Chen, W.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Liu, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Kira, Z.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Wang, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' F.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Huang, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-B.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' A Closer Look at Few-shot Classifica- tion.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In International Conference on Learning Representa- tions, 1–16.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Finn, C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Abbeel, P.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Levine, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Precup, D.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Teh, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' W.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', eds.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', Proceedings of the 34th International Conference on Machine Learning, volume 70 of Proceedings of Machine Learning Research, 1126–1135.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ganin, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ustinova, E.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ajakan, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Germain, P.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Larochelle, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Laviolette, F.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' March, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Lempitsky, V.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2016.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Domain-Adversarial Training of Neural Networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Journal of Machine Learning Research, 17(59): 1–35.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Gidaris, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Bursuc, A.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Komodakis, N.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Perez, P.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Cord, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Learning Representations by Predicting Bags of Visual Words.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR), 6926–6936.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Huang, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Yang, W.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Wang, L.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Zhou, L.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Yang, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Few-Shot Unsupervised Domain Adaptation with Image-to-Class Sparse Similarity Encoding.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Proceedings of the 29th ACM International Conference on Multimedia, MM ’21, 677–685.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' New York, NY, USA: Association for Computing Machinery.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ISBN 9781450386517.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Huang, X.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Liu, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Belongie, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Kautz, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2018.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Multimodal Unsupervised Image-to-Image Translation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Ferrari, V.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Hebert, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Sminchisescu, C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Weiss, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', eds.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', Computer Vision – ECCV 2018, 179–196.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Cham: Springer International Publishing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ISBN 978-3-030-01219- 9.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Kim, D.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Saito, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Oh, T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Plummer, B.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' A.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Sclaroff, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Saenko, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' CDS: Cross-Domain Self-supervised Pre-training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In 2021 IEEE/CVF International Conference on Computer Vision (ICCV), 9103–9112.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Li, W.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Wang, L.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Huo, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Shi, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Gao, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Luo, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Asymmetric Distribution Measure for Few-shot Learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Bessiere, C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', ed.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', Proceedings of the Twenty-Ninth Interna- tional Joint Conference on Artificial Intelligence, IJCAI-20, 2957–2963.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' International Joint Conferences on Artificial In- telligence Organization.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Main track.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Li, W.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Wang, L.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Xu, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Huo, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Gao, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Luo, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Revisiting Local Descriptor Based Image-To-Class Measure for Few-Shot Learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 7260–7268.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Long, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Cao, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Wang, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Jordan, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2015.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Learn- ing Transferable Features with Deep Adaptation Networks.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Bach, F.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Blei, D.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', eds.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', Proceedings of the 32nd In- ternational Conference on Machine Learning, volume 37 of Proceedings of Machine Learning Research, 97–105.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Lille, France: PMLR.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Luo, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Liu, P.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Guan, T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Yu, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Yang, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ad- versarial Style Mining for One-Shot Unsupervised Domain Adaptation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Larochelle, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ranzato, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Hadsell, R.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Balcan, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Lin, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', eds.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', Advances in Neural Informa- tion Processing Systems, volume 33, 20612–20623.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Curran Associates, Inc.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Peng, X.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Bai, Q.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Xia, X.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Huang, Z.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Saenko, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Wang, B.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Moment Matching for Multi-Source Do- main Adaptation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In 2019 IEEE/CVF International Confer- ence on Computer Vision (ICCV), 1406–1415.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ravi, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Larochelle, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Optimization as a Model for Few-Shot Learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In International Conference on Learning Representations.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Roy, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Siarohin, A.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Sangineto, E.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Bulo, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' R.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Sebe, N.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Ricci, E.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Unsupervised Domain Adaptation Using Feature-Whitening and Consensus Loss.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 9471–9480.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Saito, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Watanabe, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ushiku, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Harada, T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2018.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Maximum Classifier Discrepancy for Unsupervised Domain Adaptation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 3723– 3732.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Snell, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Swersky, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Zemel, R.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Prototypical Net- works for Few-shot Learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Guyon, I.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Luxburg, U.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' V.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Bengio, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Wallach, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Fergus, R.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Vishwanathan, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Garnett, R.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', eds.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', Advances in Neural Information Process- ing Systems, volume 30, 4077–4087.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Curran Associates, Inc.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Tang, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Chen, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Jia, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Unsupervised Domain Adaptation via Structurally Regularized Deep Clustering.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition (CVPR).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Tseng, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Lee, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Huang, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-B.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Yang, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Cross-Domain Few-Shot Classification via Learned Feature-Wise Transformation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In International Conference on Learning Representations.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Tzeng, E.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Hoffman, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Saenko, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Darrell, T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2017.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Adversarial Discriminative Domain Adaptation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In 2017 IEEE Conference on Computer Vision and Pattern Recog- nition (CVPR), 2962–2971.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Tzeng, E.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Hoffman, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Zhang, N.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Saenko, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Darrell, T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2014.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Deep Domain Confusion: Maximizing for Domain Invariance.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' CoRR, abs/1412.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='3474: 1–9.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Vinyals, O.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Blundell, C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Lillicrap, T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' kavukcuoglu, k.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Wierstra, D.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2016.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Matching Networks for One Shot Learn- ing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Lee, D.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Sugiyama, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Luxburg, U.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Guyon, I.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Garnett, R.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', eds.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', Advances in Neural Information Process- ing Systems, volume 29, 3630–3638.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Curran Associates, Inc.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Yang, W.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Yang, C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Huang, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Wang, L.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Yang, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2022.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Few-shot Unsupervised Domain Adaptation via Meta Learning.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In IEEE International Conference on Multimedia and Expo (ICME).' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Ye, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ;' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Hu, H.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Zhan, D.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content='-C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Sha, F.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Few-Shot Learning via Embedding Adaptation With Set-to-Set Func- tions.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR), 8805–8814.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Yue, X.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Zheng, Z.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Zhang, S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Gao, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Darrell, T.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Keutzer, K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Vincentelli, A.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' S.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2021.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Prototypical Cross-Domain Self-Supervised Learning for Few-Shot Unsupervised Do- main Adaptation.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR), 13834–13844.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Zhang, C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Cai, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Lin, G.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Shen, C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2020.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Deep- EMD: Few-Shot Image Classification With Differentiable Earth Mover’s Distance and Structured Classifiers.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 12200–12210.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Zou, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Yu, Z.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Liu, X.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Kumar, B.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' V.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' V.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Wang, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2019.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Confidence Regularized Self-Training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In 2019 IEEE/CVF International Conference on Computer Vision (ICCV), 5981–5990.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Zou, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Yu, Z.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Vijaya Kumar, B.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' V.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' K.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Wang, J.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' 2018.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Unsupervised Domain Adaptation for Semantic Segmenta- tion via Class-Balanced Self-training.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' In Ferrari, V.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Hebert, M.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Sminchisescu, C.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=';' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' and Weiss, Y.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', eds.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=', Computer Vision – ECCV 2018, 297–313.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' Cham: Springer International Pub- lishing.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'} +page_content=' ISBN 978-3-030-01219-9.' metadata={'source': '/home/zjlab/wf/langchain-ChatGLM/knowledge_base/A9AzT4oBgHgl3EQf__9t/content/2301.01956v1.pdf'}