id
stringlengths 9
16
| submitter
stringlengths 1
64
⌀ | authors
stringlengths 3
62.8k
| title
stringlengths 1
382
| comments
stringlengths 1
1.15k
⌀ | journal-ref
stringlengths 1
557
⌀ | doi
stringlengths 7
153
⌀ | report-no
stringlengths 1
509
⌀ | categories
stringlengths 5
125
| license
stringclasses 9
values | abstract
stringlengths 6
6.09k
| versions
listlengths 1
187
| update_date
timestamp[s]date 2007-05-23 00:00:00
2025-04-21 00:00:00
| authors_parsed
sequencelengths 1
2.83k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2401.09594 | William Hutchins | William Hutchins, John A. Tomko, Dan M. Hirt, Saman Zare, Joseph R.
Matson, Katja Diaz-Granados, Mingze He, Thomas Pfeifer, Jiahan Li, James
Edgar, Jon-Paul Maria, Joshua D. Caldwell, Patrick E. Hopkins | Ultrafast evanescent heat transfer across solid interfaces via
hyperbolic phonon polaritons in hexagonal boron nitride | null | null | 10.1038/s41563-025-02154-5 | null | cond-mat.mes-hall physics.optics | http://creativecommons.org/licenses/by/4.0/ | The efficiency of phonon-mediated heat transport is limited by the intrinsic
atomistic properties of materials, seemingly providing an upper limit to heat
transfer in materials and across their interfaces. The typical speeds of
conductive transport, which are inherently limited by the chemical bonds and
atomic masses, dictate how quickly heat will move in solids. Given that
phonon-polaritons, or coupled phonon-photon modes, can propagate at speeds
approaching 1 percent of the speed of light - orders of magnitude faster than
transport within a pure diffusive phonon conductor - we demonstrate that
volume-confined, hyperbolic phonon-polariton(HPhP) modes supported by many
biaxial polar crystals can couple energy across solid-solid interfaces at an
order of magnitude higher rates than phonon-phonon conduction alone. Using
pump-probe thermoreflectance with a mid-infrared, tunable, probe pulse with
sub-picosecond resolution, we demonstrate remote and spectrally selective
excitation of the HPhP modes in hexagonal boron nitride in response to
radiative heating from a thermally emitting gold source. Our work demonstrates
a new avenue for interfacial heat transfer based on broadband radiative
coupling from a hot spot in a gold film to hBN HPhPs, independent of the broad
spectral mismatch between the pump(visible) and probe(mid-IR) pulses employed.
This methodology can be used to bypass the intrinsically limiting phonon-phonon
conductive pathway, thus providing an alternative means of heat transfer across
interfaces. Further, our time-resolved measurements of the temperature changes
of the HPhP modes in hBN show that through polaritonic coupling, a material can
transfer heat across and away from an interface at rates orders of magnitude
faster than diffusive phonon speeds intrinsic to the material, thus
demonstrating a pronounced thermal transport enhancement in hBN via
phonon-polariton coupling.
| [
{
"version": "v1",
"created": "Wed, 17 Jan 2024 20:49:54 GMT"
},
{
"version": "v2",
"created": "Wed, 21 Feb 2024 00:19:58 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Hutchins",
"William",
""
],
[
"Tomko",
"John A.",
""
],
[
"Hirt",
"Dan M.",
""
],
[
"Zare",
"Saman",
""
],
[
"Matson",
"Joseph R.",
""
],
[
"Diaz-Granados",
"Katja",
""
],
[
"He",
"Mingze",
""
],
[
"Pfeifer",
"Thomas",
""
],
[
"Li",
"Jiahan",
""
],
[
"Edgar",
"James",
""
],
[
"Maria",
"Jon-Paul",
""
],
[
"Caldwell",
"Joshua D.",
""
],
[
"Hopkins",
"Patrick E.",
""
]
] |
2401.14862 | \"Ozlem Ejder | \"Ozlem Ejder | Galois theory of quadratic rational functions with periodic critical
points | null | null | null | null | math.NT math.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given a number field $k$, and a quadratic rational function $f(x) \in k(x)$,
the associated arboreal representation of the absolute Galois group of $k$ is a
subgroup of the automorphism group of a regular rooted binary tree. Boston and
Jones conjectured that the image of such a representation for $f \in
\mathbb{Z}[x]$ contains a dense set of settled elements. An automorphism is
settled if the number of its orbits on the $n\text{th}$ level of the tree
remains small as $n$ goes to infinity.
In this article, we exhibit many quadratic rational functions whose
associated Arboreal Galois groups are not densely settled. These examples arise
from quadratic rational functions whose critical points lie in a single
periodic orbit. To prove our results, we present a detailed study of the
iterated monodromy groups (IMG) of $f$, which also allows us to provide a
negative answer to Jones and Levy's question regarding settled pairs.
Furthermore, we study the iterated extension $k(f^{-\infty}(t))$ generated by
adjoining to $k(t)$ all roots of $f^n(x) = t$ for $n \geq 1$ for a parameter
$t$. We call the intersection of $k(f^{-\infty}(t))$ with $\bar{k}$, the field
of constants associated with $f$. When one of the two critical points of $f$ is
the image of the other, we show that the field of constants is contained in the
cyclotomic extension of $k$ generated by all $2$-power roots of unity. In
particular, we prove the conjecture of Ejder, Kara, and Ozman regarding the
rational function $\frac{1}{(x-1)^2}$.
| [
{
"version": "v1",
"created": "Fri, 26 Jan 2024 13:42:27 GMT"
},
{
"version": "v2",
"created": "Wed, 7 Aug 2024 08:44:57 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 11:25:49 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Ejder",
"Özlem",
""
]
] |
2401.16175 | Shenyuan Ma | Shenyuan Ma and Jakub Marecek and Vyacheslav Kungurtsev and Marek
Tyburec | Truss topology design under harmonic loads: Peak power minimization with
semidefinite programming | null | Structural and Multidisciplinary Optimization, Volume 68, article
number 40, 2025 | 10.1007/s00158-025-03973-5 | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Designing lightweight yet stiff structures that can withstand vibrations is a
crucial task in structural optimization. Here, we present a novel framework for
truss topology optimization under undamped harmonic oscillations. Our approach
minimizes the peak power of the structure under harmonic loads, overcoming the
limitations of single-frequency and in-phase assumptions found in previous
methods. For this, we leverage the concept of semidefinite representable (SDr)
functions, demonstrating that while compliance readily conforms to an SDr
representation, peak power requires a derivation based on the non-negativity of
trigonometric functions. Finally, we introduce convex relaxations for the
minimization problem and provide promising computational results.
| [
{
"version": "v1",
"created": "Mon, 29 Jan 2024 14:15:57 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Feb 2025 16:54:09 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Ma",
"Shenyuan",
""
],
[
"Marecek",
"Jakub",
""
],
[
"Kungurtsev",
"Vyacheslav",
""
],
[
"Tyburec",
"Marek",
""
]
] |
2402.03592 | Ali Khajegili Mirabadi | Ali Khajegili Mirabadi, Graham Archibald, Amirali Darbandsari, Alberto
Contreras-Sanz, Ramin Ebrahim Nakhli, Maryam Asadi, Allen Zhang, C. Blake
Gilks, Peter Black, Gang Wang, Hossein Farahani, Ali Bashashati | GRASP: GRAph-Structured Pyramidal Whole Slide Image Representation | Accepted in Learning Meaningful Representations of Life (LMRL)
Workshop at ICLR 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Cancer subtyping is one of the most challenging tasks in digital pathology,
where Multiple Instance Learning (MIL) by processing gigapixel whole slide
images (WSIs) has been in the spotlight of recent research. However, MIL
approaches do not take advantage of inter- and intra-magnification information
contained in WSIs. In this work, we present GRASP, a novel lightweight
graph-structured multi-magnification framework for processing WSIs in digital
pathology. Our approach is designed to dynamically emulate the pathologist's
behavior in handling WSIs and benefits from the hierarchical structure of WSIs.
GRASP, which introduces a convergence-based node aggregation mechanism
replacing traditional pooling mechanisms, outperforms state-of-the-art methods
by a high margin in terms of balanced accuracy, while being significantly
smaller than the closest-performing state-of-the-art models in terms of the
number of parameters. Our results show that GRASP is dynamic in finding and
consulting with different magnifications for subtyping cancers, is reliable and
stable across different hyperparameters, and can generalize when using features
from different backbones. The model's behavior has been evaluated by two expert
pathologists confirming the interpretability of the model's dynamic. We also
provide a theoretical foundation, along with empirical evidence, for our work,
explaining how GRASP interacts with different magnifications and nodes in the
graph to make predictions. We believe that the strong characteristics yet
simple structure of GRASP will encourage the development of interpretable,
structure-based designs for WSI representation in digital pathology. Data and
code can be found in https://github.com/AIMLab-UBC/GRASP
| [
{
"version": "v1",
"created": "Tue, 6 Feb 2024 00:03:44 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 07:00:20 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Mirabadi",
"Ali Khajegili",
""
],
[
"Archibald",
"Graham",
""
],
[
"Darbandsari",
"Amirali",
""
],
[
"Contreras-Sanz",
"Alberto",
""
],
[
"Nakhli",
"Ramin Ebrahim",
""
],
[
"Asadi",
"Maryam",
""
],
[
"Zhang",
"Allen",
""
],
[
"Gilks",
"C. Blake",
""
],
[
"Black",
"Peter",
""
],
[
"Wang",
"Gang",
""
],
[
"Farahani",
"Hossein",
""
],
[
"Bashashati",
"Ali",
""
]
] |
2402.09235 | Zhiyuan Zheng | Zhiyuan Zheng | Some Characterizations of Weakly Uniformly Perfect Sets | null | null | null | null | math.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, the concept of weakly uniform perfectness is considered. As an
analogue of the theory of uniform perfectness, we obtain the relationships
between weakly uniform perfectness and Bergman kernel, Poincar\'e metric and
Hausdorff content. In particular, for a bounded domain $\Omega \subset
\mathbb{C}$, we show that the uniform perfectness of $\partial \Omega$ is
equivalent to $K_{\Omega}(z) \gtrsim \delta(z)^{-2}$, where $K_{\Omega}(z)$ is
the Bergman kernel of $\Omega$ and $\delta(z)$ denotes the boundary distance.
| [
{
"version": "v1",
"created": "Wed, 14 Feb 2024 15:23:40 GMT"
},
{
"version": "v2",
"created": "Mon, 22 Apr 2024 13:40:19 GMT"
},
{
"version": "v3",
"created": "Tue, 25 Jun 2024 13:12:07 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 06:55:56 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Zheng",
"Zhiyuan",
""
]
] |
2402.09527 | Muhammad Haseeb | Muhammad Haseeb, Jinkun Geng, Daniel Duclos-Cavalcanti, Ulysses
Butler, Xiyu Hao, Radhika Mittal, Srinivas Narayana, Anirudh Sivaraman | Design and Implementation of a Scalable Financial Exchange in the Public
Cloud | null | null | null | null | cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Financial exchanges are migrating to the cloud, but the best-effort nature of
the public cloud is at odds with the stringent latency requirements of
exchanges. We present Jasper, a system for meeting the networking requirements
of financial exchanges on the public cloud. Jasper uses an overlay tree to
scalably multicast market data from an exchange to ~1000 participants with low
latency (250 microseconds) and a 1-microsecond difference in data reception
time between any two participants. Jasper reuses the same tree for scalable
inbound communication (participants to exchange), augmenting it with order
pacing and a new priority queue, Limit Order Queue (LOQ), to efficiently handle
bursts of market orders. Jasper achieves better scalability and 50% lower
latency than the AWS multicast service. During bursty market activity, LOQ
nearly doubles the order processing rate.
| [
{
"version": "v1",
"created": "Wed, 14 Feb 2024 19:08:30 GMT"
},
{
"version": "v10",
"created": "Wed, 1 Jan 2025 07:17:07 GMT"
},
{
"version": "v11",
"created": "Fri, 18 Apr 2025 01:16:17 GMT"
},
{
"version": "v2",
"created": "Fri, 8 Mar 2024 22:19:43 GMT"
},
{
"version": "v3",
"created": "Tue, 12 Mar 2024 02:26:24 GMT"
},
{
"version": "v4",
"created": "Thu, 16 May 2024 17:05:44 GMT"
},
{
"version": "v5",
"created": "Sun, 2 Jun 2024 22:06:14 GMT"
},
{
"version": "v6",
"created": "Wed, 31 Jul 2024 15:54:31 GMT"
},
{
"version": "v7",
"created": "Mon, 23 Sep 2024 19:42:17 GMT"
},
{
"version": "v8",
"created": "Mon, 30 Sep 2024 19:08:36 GMT"
},
{
"version": "v9",
"created": "Tue, 17 Dec 2024 18:35:13 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Haseeb",
"Muhammad",
""
],
[
"Geng",
"Jinkun",
""
],
[
"Duclos-Cavalcanti",
"Daniel",
""
],
[
"Butler",
"Ulysses",
""
],
[
"Hao",
"Xiyu",
""
],
[
"Mittal",
"Radhika",
""
],
[
"Narayana",
"Srinivas",
""
],
[
"Sivaraman",
"Anirudh",
""
]
] |
2402.11005 | Sarath Sivaprasad | Sarath Sivaprasad, Pramod Kaushik, Sahar Abdelnabi, Mario Fritz | A Theory of LLM Sampling: Part Descriptive and Part Prescriptive | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) are increasingly utilized in autonomous
decision-making, where they sample options from vast action spaces. However,
the heuristics that guide this sampling process remain under-explored. We study
this sampling behavior and show that this underlying heuristics resembles that
of human decision-making: comprising a descriptive component (reflecting
statistical norm) and a prescriptive component (implicit ideal encoded in the
LLM) of a concept. We show that this deviation of a sample from the statistical
norm towards a prescriptive component consistently appears in concepts across
diverse real-world domains like public health, and economic trends. To further
illustrate the theory, we demonstrate that concept prototypes in LLMs are
affected by prescriptive norms, similar to the concept of normality in humans.
Through case studies and comparison with human studies, we illustrate that in
real-world applications, the shift of samples toward an ideal value in LLMs'
outputs can result in significantly biased decision-making, raising ethical
concerns.
| [
{
"version": "v1",
"created": "Fri, 16 Feb 2024 18:28:43 GMT"
},
{
"version": "v2",
"created": "Wed, 21 Feb 2024 22:02:18 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 14:01:42 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Sivaprasad",
"Sarath",
""
],
[
"Kaushik",
"Pramod",
""
],
[
"Abdelnabi",
"Sahar",
""
],
[
"Fritz",
"Mario",
""
]
] |
2402.12170 | Kuniaki Saito | Kuniaki Saito, Kihyuk Sohn, Chen-Yu Lee, Yoshitaka Ushiku | Where is the answer? Investigating Positional Bias in Language Model
Knowledge Extraction | Code is published at https://github.com/omron-sinicx/WhereIsTheAnswer | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large language models require updates to remain up-to-date or adapt to new
domains by fine-tuning them with new documents. One key is memorizing the
latest information in a way that the memorized information is extractable with
a query prompt. However, LLMs suffer from a phenomenon called perplexity curse;
despite minimizing document perplexity during fine-tuning, LLMs struggle to
extract information through a prompt sentence. In this new knowledge
acquisition and extraction, we find a very intriguing fact that LLMs can
accurately answer questions about the first sentence, but they struggle to
extract information described in the middle or end of the documents used for
fine-tuning. Our study suggests that the auto-regressive training causes this
issue; each token is prompted by reliance on all previous tokens, which hinders
the model from recalling information from training documents by question
prompts. To conduct the in-depth study, we publish both synthetic and real
datasets, enabling the evaluation of the QA performance w.r.t. the position of
the corresponding answer in a document. Our investigation shows that even a
large model suffers from the perplexity curse, but regularization such as
denoising auto-regressive loss can enhance the information extraction from
diverse positions. These findings will be (i) a key to improving knowledge
extraction from LLMs and (ii) new elements to discuss the trade-off between RAG
and fine-tuning in adapting LLMs to a new domain.
| [
{
"version": "v1",
"created": "Fri, 16 Feb 2024 06:29:16 GMT"
},
{
"version": "v2",
"created": "Thu, 23 May 2024 06:52:31 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 00:24:19 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Saito",
"Kuniaki",
""
],
[
"Sohn",
"Kihyuk",
""
],
[
"Lee",
"Chen-Yu",
""
],
[
"Ushiku",
"Yoshitaka",
""
]
] |
2402.12499 | Kim Hammar | Kim Hammar, Tao Li, Rolf Stadler, Quanyan Zhu | Automated Security Response through Online Learning with Adaptive
Conjectures | This work has been submitted to the IEEE for possible publication | null | 10.1109/TIFS.2025.3558600 | null | cs.GT cs.AI cs.CR cs.LG cs.SY eess.SY | http://creativecommons.org/licenses/by-sa/4.0/ | We study automated security response for an IT infrastructure and formulate
the interaction between an attacker and a defender as a partially observed,
non-stationary game. We relax the standard assumption that the game model is
correctly specified and consider that each player has a probabilistic
conjecture about the model, which may be misspecified in the sense that the
true model has probability 0. This formulation allows us to capture uncertainty
and misconception about the infrastructure and the intents of the players. To
learn effective game strategies online, we design Conjectural Online Learning
(COL), a novel method where a player iteratively adapts its conjecture using
Bayesian learning and updates its strategy through rollout. We prove that the
conjectures converge to best fits, and we provide a bound on the performance
improvement that rollout enables with a conjectured model. To characterize the
steady state of the game, we propose a variant of the Berk-Nash equilibrium. We
present COL through an advanced persistent threat use case. Testbed evaluations
show that COL produces effective security strategies that adapt to a changing
environment. We also find that COL enables faster convergence than current
reinforcement learning techniques.
| [
{
"version": "v1",
"created": "Mon, 19 Feb 2024 20:06:15 GMT"
},
{
"version": "v2",
"created": "Tue, 23 Jul 2024 08:50:09 GMT"
},
{
"version": "v3",
"created": "Sat, 21 Sep 2024 08:26:58 GMT"
},
{
"version": "v4",
"created": "Sun, 5 Jan 2025 11:16:19 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Hammar",
"Kim",
""
],
[
"Li",
"Tao",
""
],
[
"Stadler",
"Rolf",
""
],
[
"Zhu",
"Quanyan",
""
]
] |
2402.15242 | Ricard Ravell Rodr\'iguez | Javier Navarro, Ricard Ravell Rodr\'iguez, and Mikel Sanz | Existence of unbiased resilient estimators in discrete quantum systems | Close to the published version | Phys. Rev. Research 7, 023060 (2025) | 10.1103/PhysRevResearch.7.023060 | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | The Cram\'er-Rao bound serves as a crucial lower limit for the mean squared
error of an estimator in frequentist parameter estimation. Paradoxically, it
requires highly accurate prior knowledge of the estimated parameter for
constructing the optimal unbiased estimator. In contrast, Bhattacharyya bounds
offer a more robust estimation framework with respect to prior accuracy by
introducing additional constraints on the estimator. In this work, we examine
divergences that arise in the computation of these bounds and establish the
conditions under which they remain valid. Notably, we show that when the number
of constraints exceeds the number of measurement outcomes, an estimator with
finite variance typically does not exist. Furthermore, we systematically
investigate the properties of these bounds using paradigmatic examples,
comparing them to the Cram\'er-Rao and Bayesian approaches.
| [
{
"version": "v1",
"created": "Fri, 23 Feb 2024 10:12:35 GMT"
},
{
"version": "v2",
"created": "Fri, 1 Mar 2024 08:33:14 GMT"
},
{
"version": "v3",
"created": "Mon, 20 Jan 2025 19:18:22 GMT"
},
{
"version": "v4",
"created": "Wed, 16 Apr 2025 07:19:07 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Navarro",
"Javier",
""
],
[
"Rodríguez",
"Ricard Ravell",
""
],
[
"Sanz",
"Mikel",
""
]
] |
2402.15864 | Alexandru Dumitrescu | Alexandru Dumitrescu, Dani Korpela, Markus Heinonen, Yogesh Verma,
Valerii Iakovlev, Vikas Garg, Harri L\"ahdesm\"aki | E(3)-equivariant models cannot learn chirality: Field-based molecular
generation | ICLR 2025 | null | null | null | cs.LG physics.chem-ph q-bio.BM | http://creativecommons.org/licenses/by/4.0/ | Obtaining the desired effect of drugs is highly dependent on their molecular
geometries. Thus, the current prevailing paradigm focuses on 3D point-cloud
atom representations, utilizing graph neural network (GNN) parametrizations,
with rotational symmetries baked in via E(3) invariant layers. We prove that
such models must necessarily disregard chirality, a geometric property of the
molecules that cannot be superimposed on their mirror image by rotation and
translation. Chirality plays a key role in determining drug safety and potency.
To address this glaring issue, we introduce a novel field-based representation,
proposing reference rotations that replace rotational symmetry constraints. The
proposed model captures all molecular geometries including chirality, while
still achieving highly competitive performance with E(3)-based methods across
standard benchmarking metrics.
| [
{
"version": "v1",
"created": "Sat, 24 Feb 2024 17:13:58 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 07:16:40 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Dumitrescu",
"Alexandru",
""
],
[
"Korpela",
"Dani",
""
],
[
"Heinonen",
"Markus",
""
],
[
"Verma",
"Yogesh",
""
],
[
"Iakovlev",
"Valerii",
""
],
[
"Garg",
"Vikas",
""
],
[
"Lähdesmäki",
"Harri",
""
]
] |
2402.17962 | Ana Trujillo Negrete | Ruy Fabila-Monroy, Sergio Gerardo G\'omez-Galicia, C\'esar
Hern\'andez-Cruz, Ana Laura Trujillo-Negrete | On the Treewidth of Token and Johnson Graphs | null | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $G$ be a graph on $n$ vertices and $1 \le k \le n$ a fixed integer. The
\textit{$k$-token graph} of $G$ is the graph $F_k(G)$ whose vertex set consists
of all $k$-subsets of the vertex set of $G$, where two vertices $A$ and $B$ are
adjacent in $F_k(G)$ whenever their symmetric difference $A\triangle B$ is an
edge of $G$. In this paper we study the treewidth of $F_k(G)$ when $G$ is a
star, path, or a complete graph. We show that in the first two cases, the
treewidth is of order $\Theta(n^{k-1})$, and of order $\Theta(n^k)$ in the
third case. We conjecture that our upper bound for the treewidth of $F_k(K_n)$
is tight. This is particularly relevant since $F_k(K_n)$ is isomorphic to the
well known Johnson graph $J(n,k)$.
| [
{
"version": "v1",
"created": "Wed, 28 Feb 2024 01:02:07 GMT"
},
{
"version": "v2",
"created": "Fri, 1 Mar 2024 17:26:34 GMT"
},
{
"version": "v3",
"created": "Thu, 17 Apr 2025 22:20:03 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Fabila-Monroy",
"Ruy",
""
],
[
"Gómez-Galicia",
"Sergio Gerardo",
""
],
[
"Hernández-Cruz",
"César",
""
],
[
"Trujillo-Negrete",
"Ana Laura",
""
]
] |
2402.19239 | Niklas Rasch | Niklas Rasch, Aleksandr N. Mikheev, Thomas Gasenzer | Bogoliubov phonons in a Bose-Einstein condensate from the one-loop
perturbative renormalization group | 21 pages, 7 figures | SciPost Phys. Core 7, 066 (2024) | 10.21468/SciPostPhysCore.7.3.066 | null | cond-mat.quant-gas hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Wilson's renormalization-group approach to the weakly-interacting
single-component Bose gas is discussed within the symmetry-broken, condensate
phase. Extending upon the work by Bijlsma and Stoof [Phys. Rev. A 54, 5085
(1996), see http://doi.org/10.1103/PhysRevA.54.5085 ], wave-function
renormalization of the temporal derivative contributions to the effective
action is included in order to capture sound-like quasiparticle excitations
with wave lengths larger than the healing-length scale. By means of a suitable
rescaling scheme we achieve convergence of the coupling flows, which serve as a
means to determine the condensate depletion in accordance with Bogoliubov
theory, as well as the interaction-induced shift of the critical temperature.
| [
{
"version": "v1",
"created": "Thu, 29 Feb 2024 15:13:39 GMT"
},
{
"version": "v2",
"created": "Fri, 2 Aug 2024 11:27:20 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Rasch",
"Niklas",
""
],
[
"Mikheev",
"Aleksandr N.",
""
],
[
"Gasenzer",
"Thomas",
""
]
] |
2403.02832 | Michael Samet | Christian Bayer, Chiheb Ben Hammouda, Antonis Papapantoleon, Michael
Samet and Ra\'ul Tempone | Quasi-Monte Carlo with Domain Transformation for Efficient Fourier
Pricing of Multi-Asset Options | null | null | null | null | q-fin.CP cs.NA math.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Efficiently pricing multi-asset options poses a significant challenge in
quantitative finance. Fourier methods leverage the regularity properties of the
integrand in the Fourier domain to accurately and rapidly value options that
typically lack regularity in the physical domain. However, most of the existing
Fourier approaches face hurdles in high-dimensional settings due to the tensor
product (TP) structure of the commonly employed numerical quadrature
techniques. To overcome this difficulty, this work advocates using the
randomized quasi-MC (RQMC) quadrature to improve the scalability of Fourier
methods with high dimensions. The RQMC technique benefits from the smoothness
of the integrand and alleviates the curse of dimensionality while providing
practical error estimates. Nonetheless, the applicability of RQMC on the
unbounded domain, $\mathbb{R}^d$, requires a domain transformation to
$[0,1]^d$, which may result in singularities of the transformed integrand at
the corners of the hypercube, and hence deteriorate the performance of RQMC. To
circumvent this difficulty, we design an efficient domain transformation
procedure based on boundary growth conditions on the transformed integrand. The
proposed transformation preserves sufficient regularity of the original
integrand for fast convergence of the RQMC method. To validate our analysis, we
demonstrate the efficiency of employing RQMC with an appropriate transformation
to evaluate options in the Fourier space for various pricing models, payoffs,
and dimensions. Finally, we highlight the computational advantage of applying
RQMC over MC or TP in the Fourier domain, and over MC in the physical domain
for options with up to 15 assets.
| [
{
"version": "v1",
"created": "Tue, 5 Mar 2024 10:06:09 GMT"
},
{
"version": "v2",
"created": "Sun, 19 Jan 2025 15:31:24 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 10:03:54 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Bayer",
"Christian",
""
],
[
"Hammouda",
"Chiheb Ben",
""
],
[
"Papapantoleon",
"Antonis",
""
],
[
"Samet",
"Michael",
""
],
[
"Tempone",
"Raúl",
""
]
] |
2403.03868 | Ying Jin | Ying Jin, Zhimei Ren | Confidence on the Focal: Conformal Prediction with Selection-Conditional
Coverage | Forthcoming at Journal of the Royal Statistical Society Series B | null | null | null | stat.ME math.ST stat.ML stat.TH | http://creativecommons.org/licenses/by/4.0/ | Conformal prediction builds marginally valid prediction intervals that cover
the unknown outcome of a randomly drawn test point with a prescribed
probability. However, in practice, data-driven methods are often used to
identify specific test unit(s) of interest, requiring uncertainty
quantification tailored to these focal units. In such cases, marginally valid
conformal prediction intervals may fail to provide valid coverage for the focal
unit(s) due to selection bias. This paper presents a general framework for
constructing a prediction set with finite-sample exact coverage, conditional on
the unit being selected by a given procedure. The general form of our method
accommodates arbitrary selection rules that are invariant to the permutation of
the calibration units, and generalizes Mondrian Conformal Prediction to
multiple test units and non-equivariant classifiers. We also work out
computationally efficient implementation of our framework for a number of
realistic selection rules, including top-K selection, optimization-based
selection, selection based on conformal p-values, and selection based on
properties of preliminary conformal prediction sets. The performance of our
methods is demonstrated via applications in drug discovery and health risk
prediction.
| [
{
"version": "v1",
"created": "Wed, 6 Mar 2024 17:18:24 GMT"
},
{
"version": "v2",
"created": "Sun, 24 Mar 2024 04:07:37 GMT"
},
{
"version": "v3",
"created": "Thu, 17 Apr 2025 19:05:51 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Jin",
"Ying",
""
],
[
"Ren",
"Zhimei",
""
]
] |
2403.04724 | Georgios Leontidis | Miles Everett, Mingjun Zhong, and Georgios Leontidis | Masked Capsule Autoencoders | 15 pages, 7 figures, 5 tables - accepted at TMLR | TMLR 01/2025 - https://openreview.net/forum?id=JHxrh00W1j | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | We propose Masked Capsule Autoencoders (MCAE), the first Capsule Network that
utilises pretraining in a modern self-supervised paradigm, specifically the
masked image modelling framework. Capsule Networks have emerged as a powerful
alternative to Convolutional Neural Networks (CNNs). They have shown favourable
properties when compared to Vision Transformers (ViT), but have struggled to
effectively learn when presented with more complex data. This has led to
Capsule Network models that do not scale to modern tasks. Our proposed MCAE
model alleviates this issue by reformulating the Capsule Network to use masked
image modelling as a pretraining stage before finetuning in a supervised
manner. Across several experiments and ablations studies we demonstrate that
similarly to CNNs and ViTs, Capsule Networks can also benefit from
self-supervised pretraining, paving the way for further advancements in this
neural network domain. For instance, by pretraining on the Imagenette
dataset-consisting of 10 classes of Imagenet-sized images-we achieve
state-of-the-art results for Capsule Networks, demonstrating a 9% improvement
compared to our baseline model. Thus, we propose that Capsule Networks benefit
from and should be trained within a masked image modelling framework, using a
novel capsule decoder, to enhance a Capsule Network's performance on
realistically sized images.
| [
{
"version": "v1",
"created": "Thu, 7 Mar 2024 18:22:03 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 07:54:51 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Everett",
"Miles",
""
],
[
"Zhong",
"Mingjun",
""
],
[
"Leontidis",
"Georgios",
""
]
] |
2403.06281 | Wei-Lun Huang | Wei-Lun Huang and Kang G. Shin | ES-FUZZ: Improving the Coverage of Firmware Fuzzing with Stateful and
Adaptable MMIO Models | 15 pages, 3 figures, 4 tables | null | null | null | cs.CR | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Gray-box fuzzing is widely used for testing embedded systems (ESes).
State-of-the-art (SOTA) gray-box fuzzers test ES firmware in fully emulated
environments without real peripherals. They emulate missing peripherals to
achieve decent code coverage. Some fuzzers infer the memory-mapped I/O (MMIO)
behavior of firmware peripherals from the firmware binary. We find that these
fuzzers emulate the inferred MMIO behavior using stateless and non-adaptive
MMIO models, which perform poorly in handling ES firmware's MMIO reads to
collectively retrieve a data chunk. This leaves ample room for improving the
code coverage of these fuzzers.
We propose ES-Fuzz to improve the code coverage of each such fuzzer using
stateful MMIO models that adapt to overcome the fuzzer's coverage bottlenecks.
ES-Fuzz runs concurrently with a given fuzzer and starts a new run whenever the
fuzzer's coverage stagnates. In each run, ES-Fuzz leverages a high-coverage
test case to generate new stateful MMIO models that boost the coverage further.
We have implemented ES-Fuzz upon Fuzzware and evaluated it with 24 popular ES
firmware. ES-Fuzz is shown to enhance Fuzzware's coverage by up to 54% in 11 of
them and trigger additional bugs in 5 of them without hurting the coverage in
the remainder. ES-Fuzz's MMIO models are shown to describe a wide range of
MMIO-retrieved data chunks and the firmware's usage of the same data chunk in
various contexts.
| [
{
"version": "v1",
"created": "Sun, 10 Mar 2024 18:19:40 GMT"
},
{
"version": "v2",
"created": "Sat, 14 Sep 2024 03:00:20 GMT"
},
{
"version": "v3",
"created": "Thu, 17 Apr 2025 23:26:26 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Huang",
"Wei-Lun",
""
],
[
"Shin",
"Kang G.",
""
]
] |
2403.06813 | Georgios Leontidis | Mohammad Alkhalefi, Georgios Leontidis, and Mingjun Zhong | LeOCLR: Leveraging Original Images for Contrastive Learning of Visual
Representations | 15 pages, 5 figures, 9 tables - accepted at TMLR 10/2024; V4
corrected some typos in the references | TMLR; 2024 | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Contrastive instance discrimination methods outperform supervised learning in
downstream tasks such as image classification and object detection. However,
these methods rely heavily on data augmentation during representation learning,
which can lead to suboptimal results if not implemented carefully. A common
augmentation technique in contrastive learning is random cropping followed by
resizing. This can degrade the quality of representation learning when the two
random crops contain distinct semantic content. To tackle this issue, we
introduce LeOCLR (Leveraging Original Images for Contrastive Learning of Visual
Representations), a framework that employs a novel instance discrimination
approach and an adapted loss function. This method prevents the loss of
important semantic features caused by mapping different object parts during
representation learning. Our experiments demonstrate that LeOCLR consistently
improves representation learning across various datasets, outperforming
baseline models. For instance, LeOCLR surpasses MoCo-v2 by 5.1% on ImageNet-1K
in linear evaluation and outperforms several other methods on transfer learning
and object detection tasks.
| [
{
"version": "v1",
"created": "Mon, 11 Mar 2024 15:33:32 GMT"
},
{
"version": "v2",
"created": "Thu, 18 Jul 2024 18:55:51 GMT"
},
{
"version": "v3",
"created": "Tue, 15 Oct 2024 15:52:15 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 11:46:18 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Alkhalefi",
"Mohammad",
""
],
[
"Leontidis",
"Georgios",
""
],
[
"Zhong",
"Mingjun",
""
]
] |
2403.07579 | Lior Arbel | Lior Arbel, Ishwarya Ananthabhotla, Zamir Ben-Hur, David Lou Alon,
Boaz Rafaely | On HRTF Notch Frequency Prediction Using Anthropometric Features and
Neural Networks | null | null | 10.1109/ICASSP48485.2024.10447610 | null | eess.AS | http://creativecommons.org/licenses/by/4.0/ | High fidelity spatial audio often performs better when produced using a
personalized head-related transfer function (HRTF). However, the direct
acquisition of HRTFs is cumbersome and requires specialized equipment. Thus,
many personalization methods estimate HRTF features from easily obtained
anthropometric features of the pinna, head, and torso. The first HRTF notch
frequency (N1) is known to be a dominant feature in elevation localization, and
thus a useful feature for HRTF personalization. This paper describes the
prediction of N1 frequency from pinna anthropometry using a neural model.
Prediction is performed separately on three databases, both simulated and
measured, and then by domain mixing in-between the databases. The model
successfully predicts N1 frequency for individual databases and by domain
mixing between some databases. Prediction errors are better or comparable to
those previously reported, showing significant improvement when acquired over a
large database and with a larger output range.
| [
{
"version": "v1",
"created": "Tue, 12 Mar 2024 12:07:56 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Arbel",
"Lior",
""
],
[
"Ananthabhotla",
"Ishwarya",
""
],
[
"Ben-Hur",
"Zamir",
""
],
[
"Alon",
"David Lou",
""
],
[
"Rafaely",
"Boaz",
""
]
] |
2403.08857 | Minbin Huang | Minbin Huang, Yanxin Long, Xinchi Deng, Ruihang Chu, Jiangfeng Xiong,
Xiaodan Liang, Hong Cheng, Qinglin Lu, Wei Liu | DialogGen: Multi-modal Interactive Dialogue System for Multi-turn
Text-to-Image Generation | Project page: https://hunyuan-dialoggen.github.io/. Accepted to
NAACL2025 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Text-to-image (T2I) generation models have significantly advanced in recent
years. However, effective interaction with these models is challenging for
average users due to the need for specialized prompt engineering knowledge and
the inability to perform multi-turn image generation, hindering a dynamic and
iterative creation process. Recent attempts have tried to equip Multi-modal
Large Language Models (MLLMs) with T2I models to bring the user's natural
language instructions into reality. Hence, the output modality of MLLMs is
extended, and the multi-turn generation quality of T2I models is enhanced
thanks to the strong multi-modal comprehension ability of MLLMs. However, many
of these works face challenges in identifying correct output modalities and
generating coherent images accordingly as the number of output modalities
increases and the conversations go deeper. Therefore, we propose DialogGen, an
effective pipeline to align off-the-shelf MLLMs and T2I models to build a
Multi-modal Interactive Dialogue System (MIDS) for multi-turn Text-to-Image
generation. It is composed of drawing prompt alignment, careful training data
curation, and error correction. Moreover, as the field of MIDS flourishes,
comprehensive benchmarks are urgently needed to evaluate MIDS fairly in terms
of output modality correctness and multi-modal output coherence. To address
this issue, we introduce the Multi-modal Dialogue Benchmark (DialogBen), a
comprehensive bilingual benchmark designed to assess the ability of MLLMs to
generate accurate and coherent multi-modal content that supports image editing.
It contains two evaluation metrics to measure the model's ability to switch
modalities and the coherence of the output images. Our extensive experiments on
DialogBen and user study demonstrate the effectiveness of DialogGen compared
with other State-of-the-Art models.
| [
{
"version": "v1",
"created": "Wed, 13 Mar 2024 18:00:01 GMT"
},
{
"version": "v2",
"created": "Wed, 3 Jul 2024 13:09:53 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 11:51:22 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Huang",
"Minbin",
""
],
[
"Long",
"Yanxin",
""
],
[
"Deng",
"Xinchi",
""
],
[
"Chu",
"Ruihang",
""
],
[
"Xiong",
"Jiangfeng",
""
],
[
"Liang",
"Xiaodan",
""
],
[
"Cheng",
"Hong",
""
],
[
"Lu",
"Qinglin",
""
],
[
"Liu",
"Wei",
""
]
] |
2403.13463 | Raymond Cheng | Raymond Cheng, Alexander Perry, Xiaolei Zhao | Derived categories of quartic double fivefolds | 23 pages, accepted version, comments always welcome! | null | null | null | math.AG | http://creativecommons.org/licenses/by/4.0/ | We construct singular quartic double fivefolds whose Kuznetsov component
admits a crepant categorical resolution of singularities by a twisted
Calabi--Yau threefold. We also construct rational specializations of these
fivefolds where such a resolution exists without a twist. This confirms an
instance of a higher-dimensional version of Kuznetsov's rationality conjecture,
and of a noncommutative version of Reid's fantasy on the connectedness of the
moduli of Calabi--Yau threefolds.
| [
{
"version": "v1",
"created": "Wed, 20 Mar 2024 10:16:05 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 13:14:05 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Cheng",
"Raymond",
""
],
[
"Perry",
"Alexander",
""
],
[
"Zhao",
"Xiaolei",
""
]
] |
2403.17309 | Kenji Homma | Kenji Homma, Satoshi Morita, and Naoki Kawashima | Tensor network renormalization approach to antiferromagnetic 6-state
clock model on the Union Jack lattice | 12 pages, 8 figures | null | 10.1103/PhysRevB.111.134427 | null | cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Using the nuclear norm regularization techniques on tensor network
renormalization algorithm, we study the phase diagram, the critical behavior
and the duality property of the antiferromagnetic 6-state clock model on the
Union Jack lattice. We find that this model undergoes multiple phase
transitions; there is the Berezinskii-Kosterlitz-Thouless, $Z_{6}$ symmetry
breaking and chiral transition with decreasing temperature. Furthermore, we
provide convincing numerical evidence that its quasi-long range order is well
explained by the compactified boson conformal field theory (CFT) and the chiral
transition is in perfect agreement with the Ising CFT, including central
charge, scaling dimension spectrum, and operator product expansion
coefficients.
| [
{
"version": "v1",
"created": "Tue, 26 Mar 2024 01:33:49 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Homma",
"Kenji",
""
],
[
"Morita",
"Satoshi",
""
],
[
"Kawashima",
"Naoki",
""
]
] |
2403.19419 | Jakub Marecek | Andrii Kliachkin and Eleni Psaroudaki and Jakub Marecek and Dimitris
Fotakis | Fairness in Ranking: Robustness through Randomization without the
Protected Attribute | null | 2024 IEEE 40th International Conference on Data Engineering
Workshops | 10.1109/ICDEW61823.2024.00032 | null | cs.LG cs.AI cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | There has been great interest in fairness in machine learning, especially in
relation to classification problems. In ranking-related problems, such as in
online advertising, recommender systems, and HR automation, much work on
fairness remains to be done. Two complications arise: first, the protected
attribute may not be available in many applications. Second, there are multiple
measures of fairness of rankings, and optimization-based methods utilizing a
single measure of fairness of rankings may produce rankings that are unfair
with respect to other measures. In this work, we propose a randomized method
for post-processing rankings, which do not require the availability of the
protected attribute. In an extensive numerical study, we show the robustness of
our methods with respect to P-Fairness and effectiveness with respect to
Normalized Discounted Cumulative Gain (NDCG) from the baseline ranking,
improving on previously proposed methods.
| [
{
"version": "v1",
"created": "Thu, 28 Mar 2024 13:50:24 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Kliachkin",
"Andrii",
""
],
[
"Psaroudaki",
"Eleni",
""
],
[
"Marecek",
"Jakub",
""
],
[
"Fotakis",
"Dimitris",
""
]
] |
2404.01202 | Jef Pauwels | Tristan Le Roy-Deloison, Edwin Peter Lobo, Jef Pauwels, Stefano
Pironio | Device-independent quantum key distribution based on routed Bell tests | 14 + N pages, 16 figures, 1 table | PRX Quantum 6, 020311 (April 2025) | 10.1103/PRXQuantum.6.020311 | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Photon losses are the main obstacle to fully photonic implementations of
device-independent quantum key distribution (DIQKD). Motivated by recent work
showing that routed Bell scenarios offer increased robustness to detection
inefficiencies for the certification of long-range quantum correlations, we
investigate DIQKD protocols based on a routed setup. In these protocols, in
some of the test rounds, photons from the source are routed by an actively
controlled switch to a nearby test device instead of the distant one. We show
how to analyze the security of these protocols and compute lower bounds on the
key rates using non-commutative polynomial optimization and the
Brown-Fawzi-Fazwi method. We determine lower bounds on the asymptotic key rates
of several simple two-qubit routed DIQKD protocols based on CHSH or BB84
correlations and compare their performance to standard protocols. We find that
in an ideal case routed DIQKD protocols can significantly improve detection
efficiency requirements, by up to $\sim 30\%$, compared to their non-routed
counterparts. Notably, the routed BB84 protocol achieves a positive key rate
with a detection efficiency as low as $50\%$ for the distant device, the
minimal threshold for any QKD protocol featuring two untrusted measurements.
However, the advantages we find are highly sensitive to noise and losses
affecting the short-range correlations involving the additional test device.
| [
{
"version": "v1",
"created": "Mon, 1 Apr 2024 15:59:09 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Roy-Deloison",
"Tristan Le",
""
],
[
"Lobo",
"Edwin Peter",
""
],
[
"Pauwels",
"Jef",
""
],
[
"Pironio",
"Stefano",
""
]
] |
2404.02151 | Maksym Andriushchenko | Maksym Andriushchenko and Francesco Croce and Nicolas Flammarion | Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks | Accepted at ICLR 2025. Updates in the v3: GPT-4o and Claude 3.5
Sonnet results, improved writing. Updates in the v2: more models (Llama3,
Phi-3, Nemotron-4-340B), jailbreak artifacts for all attacks are available,
evaluation with different judges (Llama-3-70B and Llama Guard 2), more
experiments (convergence plots, ablation on the suffix length for random
search), examples of jailbroken generation | null | null | null | cs.CR cs.AI cs.LG stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that even the most recent safety-aligned LLMs are not robust to
simple adaptive jailbreaking attacks. First, we demonstrate how to successfully
leverage access to logprobs for jailbreaking: we initially design an
adversarial prompt template (sometimes adapted to the target LLM), and then we
apply random search on a suffix to maximize a target logprob (e.g., of the
token "Sure"), potentially with multiple restarts. In this way, we achieve 100%
attack success rate -- according to GPT-4 as a judge -- on Vicuna-13B,
Mistral-7B, Phi-3-Mini, Nemotron-4-340B, Llama-2-Chat-7B/13B/70B,
Llama-3-Instruct-8B, Gemma-7B, GPT-3.5, GPT-4o, and R2D2 from HarmBench that
was adversarially trained against the GCG attack. We also show how to jailbreak
all Claude models -- that do not expose logprobs -- via either a transfer or
prefilling attack with a 100% success rate. In addition, we show how to use
random search on a restricted set of tokens for finding trojan strings in
poisoned models -- a task that shares many similarities with jailbreaking --
which is the algorithm that brought us the first place in the SaTML'24 Trojan
Detection Competition. The common theme behind these attacks is that adaptivity
is crucial: different models are vulnerable to different prompting templates
(e.g., R2D2 is very sensitive to in-context learning prompts), some models have
unique vulnerabilities based on their APIs (e.g., prefilling for Claude), and
in some settings, it is crucial to restrict the token search space based on
prior knowledge (e.g., for trojan detection). For reproducibility purposes, we
provide the code, logs, and jailbreak artifacts in the JailbreakBench format at
https://github.com/tml-epfl/llm-adaptive-attacks.
| [
{
"version": "v1",
"created": "Tue, 2 Apr 2024 17:58:27 GMT"
},
{
"version": "v2",
"created": "Tue, 18 Jun 2024 17:29:04 GMT"
},
{
"version": "v3",
"created": "Mon, 7 Oct 2024 16:35:15 GMT"
},
{
"version": "v4",
"created": "Thu, 17 Apr 2025 18:55:45 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Andriushchenko",
"Maksym",
""
],
[
"Croce",
"Francesco",
""
],
[
"Flammarion",
"Nicolas",
""
]
] |
2404.04462 | Caelan Atamanchuk | Caelan Atamanchuk and Luc Devroye and Gabor Lugosi | On the size of temporal cliques in subcritical random temporal graphs | null | null | null | null | math.PR | http://creativecommons.org/licenses/by/4.0/ | A \emph{random temporal graph} is an Erd\H{o}s-R\'enyi random graph $G(n,p)$,
together with a random ordering of its edges. A path in the graph is called
\emph{increasing} if the edges on the path appear in increasing order. A set
$S$ of vertices forms a \emph{temporal clique} if for all $u,v \in S$, there is
an increasing path from $u$ to $v$. \cite{Becker2023} proved that if $p=c\log
n/n$ for $c>1$, then, with high probability, there is a temporal clique of size
$n-o(n)$. On the other hand, for $c<1$, with high probability, the largest
temporal clique is of size $o(n)$. In this note we improve the latter bound by
showing that, for $c<1$, the largest temporal clique is of \emph{constant} size
with high probability.
| [
{
"version": "v1",
"created": "Sat, 6 Apr 2024 01:08:49 GMT"
},
{
"version": "v2",
"created": "Sun, 28 Apr 2024 15:40:56 GMT"
},
{
"version": "v3",
"created": "Thu, 17 Apr 2025 21:42:55 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Atamanchuk",
"Caelan",
""
],
[
"Devroye",
"Luc",
""
],
[
"Lugosi",
"Gabor",
""
]
] |
2404.06549 | Anna Kuzina | Haotian Chen, Anna Kuzina, Babak Esmaeili, Jakub M Tomczak | Variational Stochastic Gradient Descent for Deep Neural Networks | null | null | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | Current state-of-the-art optimizers are adaptive gradient-based optimization
methods such as Adam. Recently, there has been an increasing interest in
formulating gradient-based optimizers in a probabilistic framework for better
modeling the uncertainty of the gradients. Here, we propose to combine both
approaches, resulting in the Variational Stochastic Gradient Descent (VSGD)
optimizer. We model gradient updates as a probabilistic model and utilize
stochastic variational inference (SVI) to derive an efficient and effective
update rule. Further, we show how our VSGD method relates to other adaptive
gradient-based optimizers like Adam. Lastly, we carry out experiments on two
image classification datasets and four deep neural network architectures, where
we show that VSGD outperforms Adam and SGD.
| [
{
"version": "v1",
"created": "Tue, 9 Apr 2024 18:02:01 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 10:32:24 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Chen",
"Haotian",
""
],
[
"Kuzina",
"Anna",
""
],
[
"Esmaeili",
"Babak",
""
],
[
"Tomczak",
"Jakub M",
""
]
] |
2404.06960 | Etienne Tanr\'e | Lo\"ic B\'ethencourt, R\'emi Catellier and Etienne Tanr\'e | Brownian particles controlled by their occupation measure | null | null | 10.1137/24M1656220 | null | math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this article, we study a finite horizon linear-quadratic stochastic
control problem for Brownian particles, where the cost functions depend on the
state and the occupation measure of the particles. To address this problem, we
develop an It\^o formula for the flow of occupation measure, which enables us
to derive the associated Hamilton-Jacobi-Bellman equation. Then, thanks to a
Feynman-Kac formula and the Bou\'e-Dupuis formula, we construct an optimal
strategy and an optimal trajectory. Finally, we illustrate our result when the
cost-function is the volume of the sausage associated to the particles.
| [
{
"version": "v1",
"created": "Wed, 10 Apr 2024 12:19:06 GMT"
},
{
"version": "v2",
"created": "Wed, 29 May 2024 13:40:16 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Béthencourt",
"Loïc",
""
],
[
"Catellier",
"Rémi",
""
],
[
"Tanré",
"Etienne",
""
]
] |
2404.08121 | Kisun Lee | May Cai, Kisun Lee, Josephine Yu | The Tropical Variety of Symmetric Rank 2 Matrices | 21 pages, 8 figures, Version to appear in Linear Algebra and its
Applications | null | null | null | math.CO math.AG | http://creativecommons.org/licenses/by/4.0/ | We study the tropicalization of the variety of symmetric rank two matrices.
Analogously to the result of Markwig and Yu for general tropical rank two
matrices, we show that it has a simplicial complex structure as the space of
symmetric bicolored trees and that this simplicial complex is shellable. We
also discuss some matroid structures arising from this space and present
generating functions for the number of symmetric bicolored trees.
| [
{
"version": "v1",
"created": "Thu, 11 Apr 2024 20:56:50 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Apr 2025 18:52:14 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Cai",
"May",
""
],
[
"Lee",
"Kisun",
""
],
[
"Yu",
"Josephine",
""
]
] |
2404.14398 | P\'eter \'Agoston | P\'eter \'Agoston | A lower bound on the number of colours needed to nicely colour a sphere | The result was presented at CCCG 2020. The present paper is a revised
version of the paper in the conference proceedings | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Hadwiger--Nelson problem is about determining the chromatic number of the
plane (CNP), defined as the minimum number of colours needed to colour the
plane so that no two points of distance 1 have the same colour. In this paper
we investigate a related problem for spheres and we use a few natural
restrictions on the colouring. Thomassen showed that with these restrictions,
the chromatic number of all manifolds satisfying certain properties (including
the plane and all spheres with a large enough radius) is at least 7. We prove
that with these restrictions, the chromatic number of any sphere with a large
enough radius is at least 8. This also gives a new lower bound for the minimum
colours needed for colouring the 3-dimensional space with the same
restrictions.
| [
{
"version": "v1",
"created": "Mon, 22 Apr 2024 17:57:46 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 17:55:55 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Ágoston",
"Péter",
""
]
] |
2404.17341 | Raymond Cheng | Raymond Cheng | Free curves in Fano hypersurfaces must have high degree | 5 pages, accepted version, comments always welcome! | null | 10.1090/proc/17206 | null | math.AG | http://creativecommons.org/licenses/by/4.0/ | The purpose of this note is to show that the minimal $e$ for which every
smooth Fano hypersurface of dimension $n$ contains a free rational curve of
degree at most $e$ cannot be bounded by a linear function in $n$ when the base
field has positive characteristic. This is done by providing a super-linear
bound on the minimal possible degree of a free curve in certain Fermat
hypersurfaces.
| [
{
"version": "v1",
"created": "Fri, 26 Apr 2024 11:43:43 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 13:21:17 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Cheng",
"Raymond",
""
]
] |
2404.18693 | Philippe Gaucher | Philippe Gaucher | Natural homotopy of multipointed d-spaces | 41 pages, 9 figures | null | null | null | math.AT math.CT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We identify Grandis' directed spaces as a full reflective subcategory of the
category of multipointed $d$-spaces. When the multipointed $d$-space realizes a
precubical set, its reflection coincides with the standard realization of the
precubical set as a directed space. The reflection enables us to extend the
construction of the natural system of topological spaces in Baues-Wirsching's
sense from directed spaces to multipointed $d$-spaces. In the case of a
cellular multipointed $d$-space, there is a discrete version of this natural
system which is proved to be bisimilar up to homotopy. We also prove that these
constructions are invariant up to homotopy under globular subdivision. These
results are the globular analogue of Dubut's results. Finally, we point the
apparent incompatibility between the notion of bisimilar natural systems and
the q-model structure of multipointed $d$-spaces and we give some suggestions
for future works.
| [
{
"version": "v1",
"created": "Mon, 29 Apr 2024 13:42:15 GMT"
},
{
"version": "v2",
"created": "Thu, 16 May 2024 11:59:31 GMT"
},
{
"version": "v3",
"created": "Tue, 3 Sep 2024 13:45:00 GMT"
},
{
"version": "v4",
"created": "Tue, 12 Nov 2024 09:21:44 GMT"
},
{
"version": "v5",
"created": "Fri, 18 Apr 2025 12:45:44 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Gaucher",
"Philippe",
""
]
] |
2405.00701 | Rihito Sakurai | Rihito Sakurai, Haruto Takahashi and Koichi Miyamoto | Learning parameter dependence for Fourier-based option pricing with
tensor trains | null | null | null | null | q-fin.CP quant-ph | http://creativecommons.org/licenses/by/4.0/ | A long-standing issue in mathematical finance is the speed-up of option
pricing, especially for multi-asset options. A recent study has proposed to use
tensor train learning algorithms to speed up Fourier transform (FT)-based
option pricing, utilizing the ability of tensor trains to compress
high-dimensional tensors. Another usage of the tensor train is to compress
functions, including their parameter dependence. Here, we propose a pricing
method, where, by a tensor train learning algorithm, we build tensor trains
that approximate functions appearing in FT-based option pricing with their
parameter dependence and efficiently calculate the option price for the varying
input parameters. As a benchmark test, we run the proposed method to price a
multi-asset option for the various values of volatilities and present asset
prices. We show that, in the tested cases involving up to 11 assets, the
proposed method outperforms Monte Carlo-based option pricing with $10^6$ paths
in terms of computational complexity while keeping better accuracy.
| [
{
"version": "v1",
"created": "Wed, 17 Apr 2024 01:57:19 GMT"
},
{
"version": "v2",
"created": "Fri, 3 May 2024 01:12:28 GMT"
},
{
"version": "v3",
"created": "Tue, 7 May 2024 05:29:49 GMT"
},
{
"version": "v4",
"created": "Mon, 13 May 2024 00:54:42 GMT"
},
{
"version": "v5",
"created": "Tue, 25 Jun 2024 02:02:44 GMT"
},
{
"version": "v6",
"created": "Wed, 9 Oct 2024 10:28:00 GMT"
},
{
"version": "v7",
"created": "Thu, 31 Oct 2024 09:07:34 GMT"
},
{
"version": "v8",
"created": "Fri, 18 Apr 2025 01:11:37 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Sakurai",
"Rihito",
""
],
[
"Takahashi",
"Haruto",
""
],
[
"Miyamoto",
"Koichi",
""
]
] |
2405.00893 | Xiaobin Ding | Cunqiang Wu, Xiaobin Ding, Qi Guo, Ke Yao, Jialin Liu, Yunqing Fu, and
Chenzhong Dong | Stepwise ionization of Mo$^{14+}$ ions in EBIT: The importance of the
metastable level | null | Physical Review A, 111, 042818, 2025 | 10.1103/physreva.111.042818 | null | physics.atom-ph | http://creativecommons.org/licenses/by/4.0/ | The visible spectrum of Mo$^{15+}$ ions was measured using a high-temperature
superconducting electron-beam ion trap at the Shanghai EBIT Laboratory, with an
electron beam energy $E_{e}$=400 eV, significantly lower than the ionization
potential (IP=544.0 eV) of Mo$^{14+}$ ions in the ground state. To expound on
the experiment, the energy level structure, radiative transition properties,
electron-impact excitation, and electron-impact ionization cross section for
both the ground state and low-lying excited state of the Mo$^{14+}$ ions were
calculated using Dirac-Fock-Slater method with a local central potential and
distorted wave approximation. The results demonstrated reasonable agreement
with both available experimental and theoretical data. Through an analysis of
the related atomic processes of Mo$^{14+}$ ion, a scenario involving the
stepwise ionization of the metastable state 3p$^{6}$3d$^{9}$4s was proposed to
explain the presence of the Mo$^{15+}$ ions with a lower energy of the incident
electron. Finally, the significance of the metastable levels in ionizing
Mo$^{14+}$ ions is highlighted.
| [
{
"version": "v1",
"created": "Wed, 1 May 2024 22:36:26 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Wu",
"Cunqiang",
""
],
[
"Ding",
"Xiaobin",
""
],
[
"Guo",
"Qi",
""
],
[
"Yao",
"Ke",
""
],
[
"Liu",
"Jialin",
""
],
[
"Fu",
"Yunqing",
""
],
[
"Dong",
"Chenzhong",
""
]
] |
2405.00998 | Yuhang Huang | Yuhang Huang and SHilong Zou and Xinwang Liu and Kai Xu | Part-aware Shape Generation with Latent 3D Diffusion of Neural Voxel
Fields | This paper is accepted by TVCG | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a novel latent 3D diffusion model for the generation of
neural voxel fields, aiming to achieve accurate part-aware structures. Compared
to existing methods, there are two key designs to ensure high-quality and
accurate part-aware generation. On one hand, we introduce a latent 3D diffusion
process for neural voxel fields, enabling generation at significantly higher
resolutions that can accurately capture rich textural and geometric details. On
the other hand, a part-aware shape decoder is introduced to integrate the part
codes into the neural voxel fields, guiding the accurate part decomposition and
producing high-quality rendering results. Through extensive experimentation and
comparisons with state-of-the-art methods, we evaluate our approach across four
different classes of data. The results demonstrate the superior generative
capabilities of our proposed method in part-aware shape generation,
outperforming existing state-of-the-art methods.
| [
{
"version": "v1",
"created": "Thu, 2 May 2024 04:31:17 GMT"
},
{
"version": "v2",
"created": "Thu, 9 May 2024 12:18:48 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Jun 2024 08:49:50 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 14:07:56 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Huang",
"Yuhang",
""
],
[
"Zou",
"SHilong",
""
],
[
"Liu",
"Xinwang",
""
],
[
"Xu",
"Kai",
""
]
] |
2405.02079 | Gabriel Freedman | Gabriel Freedman, Adam Dejl, Deniz Gorur, Xiang Yin, Antonio Rago,
Francesca Toni | Argumentative Large Language Models for Explainable and Contestable
Claim Verification | 18 pages, 18 figures. Accepted as an oral presentation at AAAI 2025 | Proceedings of the AAAI Conference on Artificial Intelligence,
39(14), 14930-14939. 2025 | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by-sa/4.0/ | The profusion of knowledge encoded in large language models (LLMs) and their
ability to apply this knowledge zero-shot in a range of settings makes them
promising candidates for use in decision-making. However, they are currently
limited by their inability to provide outputs which can be faithfully explained
and effectively contested to correct mistakes. In this paper, we attempt to
reconcile these strengths and weaknesses by introducing \emph{argumentative
LLMs (ArgLLMs)}, a method for augmenting LLMs with argumentative reasoning.
Concretely, ArgLLMs construct argumentation frameworks, which then serve as the
basis for formal reasoning in support of decision-making. The interpretable
nature of these argumentation frameworks and formal reasoning means that any
decision made by ArgLLMs may be explained and contested. We evaluate ArgLLMs'
performance experimentally in comparison with state-of-the-art techniques, in
the context of the decision-making task of claim verification. We also define
novel properties to characterise contestability and assess ArgLLMs formally in
terms of these properties.
| [
{
"version": "v1",
"created": "Fri, 3 May 2024 13:12:28 GMT"
},
{
"version": "v2",
"created": "Mon, 17 Feb 2025 15:05:06 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 11:20:24 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Freedman",
"Gabriel",
""
],
[
"Dejl",
"Adam",
""
],
[
"Gorur",
"Deniz",
""
],
[
"Yin",
"Xiang",
""
],
[
"Rago",
"Antonio",
""
],
[
"Toni",
"Francesca",
""
]
] |
2405.02174 | Min Woong Ahn | Min Woong Ahn | An elementary proof that the set of exceptions to the law of large
numbers in Pierce expansions has full Hausdorff dimension | 16 pages; conclusion section added; references revised; typos
corrected | AIMS Math. 10 (2025), 6025-6039 | 10.3934/math.2025275 | null | math.NT math.CA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The digits of the Pierce expansion satisfy the law of large numbers. It is
known that the Hausdorff dimension of the set of exceptions to the law of large
numbers is 1. We provide an elementary proof of this fact by adapting Jun Wu's
method, which was originally used for Engel expansions. Our approach emphasizes
the fractal nature of exceptional sets and avoids advanced machinery, thereby
relying instead on explicit sequences and constructive techniques. Furthermore,
our method opens the possibility of extending similar analyses to other real
number representation systems, such as the Engel, L\"uroth, and Sylvester
expansions, thus paving the way for further explorations in metric number
theory and fractal geometry.
| [
{
"version": "v1",
"created": "Fri, 3 May 2024 15:25:06 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 13:28:14 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Ahn",
"Min Woong",
""
]
] |
2405.08817 | Subir Sachdev | Pietro M. Bonetti and Maine Christos and Subir Sachdev | Quantum oscillations in the hole-doped cuprates and the confinement of
spinons | v5 Added acknowledgement to Moore Foundation | Proceedings of the National Academy of Sciences 121, e2418633121
(2024) | 10.1073/pnas.2418633121 | null | cond-mat.str-el cond-mat.supr-con | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A long standing problem in the study of the under-hole-doped cuprates has
been the description of the Fermi surfaces underlying the high magnetic field
quantum oscillations, and their connection to the higher temperature pseudogap
metal. Harrison and Sebastian (arXiv:1103.4181) proposed that the pseudogap
`Fermi arcs' are reconstructed into an electron pocket by field-induced charge
density wave order. But computations on such a model (Zhang and Mei,
arXiv:1411.2098) show an unobserved additional oscillation frequency from a
Fermi surface arising from the backsides of the hole pockets completing the
Fermi arcs. We describe a transition from a fractionalized Fermi liquid (FL*)
model of the pseudogap metal, to a metal with bi-directional charge density
wave order without fractionalization. We show that the confinement of the
fermionic spinon excitations of the FL* across this transition can eliminate
the unobserved oscillation frequency.
| [
{
"version": "v1",
"created": "Tue, 14 May 2024 18:14:05 GMT"
},
{
"version": "v2",
"created": "Sat, 25 May 2024 15:20:43 GMT"
},
{
"version": "v3",
"created": "Tue, 23 Jul 2024 14:58:04 GMT"
},
{
"version": "v4",
"created": "Mon, 4 Nov 2024 16:40:19 GMT"
},
{
"version": "v5",
"created": "Fri, 18 Apr 2025 17:40:24 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Bonetti",
"Pietro M.",
""
],
[
"Christos",
"Maine",
""
],
[
"Sachdev",
"Subir",
""
]
] |
2405.10763 | Giovanni Piccioli | Rayan Harfouche, Giovanni Piccioli, Lenka Zdeborov\'a | Integer Traffic Assignment Problem: Algorithms and Insights on Random
Graphs | 37 pages, 15 figures | Physical Review E 111.1 (2025): 014316 | 10.1103/PhysRevE.111.014316 | null | cond-mat.dis-nn cs.DM math.OC stat.CO | http://creativecommons.org/licenses/by/4.0/ | Path optimization is a fundamental concern across various real-world
scenarios, ranging from traffic congestion issues to efficient data routing
over the internet. The Traffic Assignment Problem (TAP) is a classic continuous
optimization problem in this field. This study considers the Integer Traffic
Assignment Problem (ITAP), a discrete variant of TAP. ITAP involves determining
optimal routes for commuters in a city represented by a graph, aiming to
minimize congestion while adhering to integer flow constraints on paths. This
restriction makes ITAP an NP-hard problem. While conventional TAP prioritizes
repulsive interactions to minimize congestion, this work also explores the case
of attractive interactions, related to minimizing the number of occupied edges.
We present and evaluate multiple algorithms to address ITAP, including a
message passing algorithm, a greedy approach, simulated annealing, and
relaxation of ITAP to TAP. Inspired by studies of random ensembles in the
large-size limit in statistical physics, comparisons between these algorithms
are conducted on large sparse random regular graphs with a random set of
origin-destination pairs. Our results indicate that while the simplest greedy
algorithm performs competitively in the repulsive scenario, in the attractive
case the message-passing-based algorithm and simulated annealing demonstrate
superiority. We then investigate the relationship between TAP and ITAP in the
repulsive case. We find that, as the number of paths increases, the solution of
TAP converges toward that of ITAP, and we investigate the speed of this
convergence. Depending on the number of paths, our analysis leads us to
identify two scaling regimes: in one the average flow per edge is of order one,
and in another the number of paths scales quadratically with the size of the
graph, in which case the continuous relaxation solves the integer problem
closely.
| [
{
"version": "v1",
"created": "Fri, 17 May 2024 13:21:23 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Harfouche",
"Rayan",
""
],
[
"Piccioli",
"Giovanni",
""
],
[
"Zdeborová",
"Lenka",
""
]
] |
2405.13341 | Takeshi Kato | Takeshi Kato, Yosuke Tanabe, Mohammad Rezoanul Hoque | Wealth inequality and utility: Effect evaluation of redistribution and
consumption morals using the macro-econophysical coupled approach | 27 pages, 6 figures | null | null | null | econ.GN cs.MA physics.soc-ph q-fin.EC | http://creativecommons.org/licenses/by/4.0/ | Reducing wealth inequality and increasing utility are critical issues. This
study reveals the effects of redistribution and consumption morals on wealth
inequality and utility. To this end, we present a novel approach that couples
the dynamic model of capital, consumption, and utility in macroeconomics with
the interaction model of joint business and redistribution in econophysics.
With this approach, we calculate the capital (wealth), the utility based on
consumption, and the Gini index of these inequality using redistribution and
consumption thresholds as moral parameters. The results show that:
under-redistribution and waste exacerbate inequality; conversely,
over-redistribution and stinginess reduce utility; and a balanced moderate
moral leads to achieve both reduced inequality and increased utility. These
findings provide renewed economic and numerical support for the moral
importance known from philosophy, anthropology, and religion. The revival of
redistribution and consumption morals should promote the transformation to a
human mutual-aid economy, as indicated by philosopher and anthropologist,
instead of the capitalist economy that has produced the current inequality. The
practical challenge is to implement bottom-up social business, on a foothold of
worker coops and platform cooperatives as a community against the state and the
market, with moral consensus and its operation.
| [
{
"version": "v1",
"created": "Wed, 22 May 2024 04:53:21 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 03:33:31 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Kato",
"Takeshi",
""
],
[
"Tanabe",
"Yosuke",
""
],
[
"Hoque",
"Mohammad Rezoanul",
""
]
] |
2405.13828 | Ziqiao Ma | Ziqiao Ma, Zekun Wang, Joyce Chai | Babysit A Language Model From Scratch: Interactive Language Learning by
Trials and Demonstrations | NAACL 2025 (Main) & Workshop on Large Language Models and Cognition @
ICML 2024 (Oral) | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Humans are efficient language learners and inherently social creatures. Our
language development is largely shaped by our social interactions, for example,
the demonstration and feedback from caregivers. Contrary to human language
learning, recent advancements in large language models have primarily adopted a
non-interactive training paradigm, and refined pre-trained models through
feedback afterward. In this work, we explore how corrective feedback from
interactions influences neural language acquisition from scratch through
systematically controlled experiments, assessing whether it contributes to word
learning efficiency in language models. We introduce a trial-and-demonstration
(TnD) learning framework that incorporates three distinct components: student
trials, teacher demonstrations, and a reward conditioned on language competence
at various developmental stages. Our experiments reveal that the TnD approach
accelerates word acquisition for student models of equal and smaller numbers of
parameters, and we highlight the significance of both trials and
demonstrations. We further show that the teacher's choices of words influence
students' word-specific learning efficiency, and a practice-makes-perfect
effect is evident by a strong correlation between the frequency of words in
trials and their respective learning curves. Our findings suggest that
interactive language learning, with teacher demonstrations and active trials,
can facilitate efficient word learning in language models.
| [
{
"version": "v1",
"created": "Wed, 22 May 2024 16:57:02 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 16:06:57 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Ma",
"Ziqiao",
""
],
[
"Wang",
"Zekun",
""
],
[
"Chai",
"Joyce",
""
]
] |
2405.14953 | Haoxian Chen | Haoxian Chen, Hanyang Zhao, Henry Lam, David Yao, Wenpin Tang | MallowsPO: Fine-Tune Your LLM with Preference Dispersions | null | null | null | null | cs.LG cs.AI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Direct Preference Optimization (DPO) has recently emerged as a popular
approach to improve reinforcement learning with human feedback (RLHF), leading
to better techniques to fine-tune large language models (LLM). A weakness of
DPO, however, lies in its lack of capability to characterize the diversity of
human preferences. Inspired by Mallows' theory of preference ranking, we
develop in this paper a new approach, the MallowsPO. A distinct feature of this
approach is a dispersion index, which reflects the dispersion of human
preference to prompts. We show that existing DPO models can be reduced to
special cases of this dispersion index, thus unified with MallowsPO. More
importantly, we demonstrate (empirically) how to use this dispersion index to
enhance the performance of DPO in a broad array of benchmark tasks, from
synthetic bandit selection to controllable generations and dialogues, while
maintaining great generalization capabilities. MallowsPO is also compatible
with other SOTA offline preference optimization methods, boosting nearly 2\%
extra LC win rate when used as a plugin for fine-tuning Llama3-Instruct.
| [
{
"version": "v1",
"created": "Thu, 23 May 2024 18:01:11 GMT"
},
{
"version": "v2",
"created": "Sun, 15 Sep 2024 06:57:04 GMT"
},
{
"version": "v3",
"created": "Wed, 2 Oct 2024 14:56:33 GMT"
},
{
"version": "v4",
"created": "Fri, 28 Feb 2025 22:03:32 GMT"
},
{
"version": "v5",
"created": "Thu, 17 Apr 2025 18:52:54 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Chen",
"Haoxian",
""
],
[
"Zhao",
"Hanyang",
""
],
[
"Lam",
"Henry",
""
],
[
"Yao",
"David",
""
],
[
"Tang",
"Wenpin",
""
]
] |
2405.19653 | Patrick Emami | Patrick Emami, Zhaonan Li, Saumya Sinha, Truc Nguyen | SysCaps: Language Interfaces for Simulation Surrogates of Complex
Systems | Accepted at ICLR 2025. 23 pages. Updated with final camera ready
version | null | null | null | cs.LG cs.CL cs.SY eess.SY | http://creativecommons.org/licenses/by/4.0/ | Surrogate models are used to predict the behavior of complex energy systems
that are too expensive to simulate with traditional numerical methods. Our work
introduces the use of language descriptions, which we call ``system captions''
or SysCaps, to interface with such surrogates. We argue that interacting with
surrogates through text, particularly natural language, makes these models more
accessible for both experts and non-experts. We introduce a lightweight
multimodal text and timeseries regression model and a training pipeline that
uses large language models (LLMs) to synthesize high-quality captions from
simulation metadata. Our experiments on two real-world simulators of buildings
and wind farms show that our SysCaps-augmented surrogates have better accuracy
on held-out systems than traditional methods while enjoying new generalization
abilities, such as handling semantically related descriptions of the same test
system. Additional experiments also highlight the potential of SysCaps to
unlock language-driven design space exploration and to regularize training
through prompt augmentation.
| [
{
"version": "v1",
"created": "Thu, 30 May 2024 03:12:04 GMT"
},
{
"version": "v2",
"created": "Wed, 2 Oct 2024 16:23:12 GMT"
},
{
"version": "v3",
"created": "Tue, 11 Mar 2025 00:01:37 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 16:49:44 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Emami",
"Patrick",
""
],
[
"Li",
"Zhaonan",
""
],
[
"Sinha",
"Saumya",
""
],
[
"Nguyen",
"Truc",
""
]
] |
2405.19874 | Hao Zhao | Hao Zhao, Maksym Andriushchenko, Francesco Croce, Nicolas Flammarion | Is In-Context Learning Sufficient for Instruction Following in LLMs? | Accepted at ICLR 2025. This camera-ready version v3 adds multi-turn
alignment via ICL, revisiting main results on instruct models, and simple
mechanistic study. Updates in the v2: experiment with decoding schemes,
scaling in-context alignment, ICL vs IFT for instruction following. Code at
https://github.com/tml-epfl/icl-alignment | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | In-context learning (ICL) allows LLMs to learn from examples without changing
their weights: this is a particularly promising capability for long-context
LLMs that can potentially learn from many examples. Recently, Lin et al. (2024)
proposed URIAL, a method using only three in-context examples to align base
LLMs, achieving non-trivial instruction following performance. In this work, we
show that, while effective, ICL alignment with URIAL still underperforms
compared to instruction fine-tuning on the established benchmark MT-Bench,
especially with more capable base LLMs. We then uncover the most relevant
elements for successful in-context alignment, finding the crucial role of the
decoding parameters. Based on these insights, we show that the approach of
URIAL can indeed be improved by adding high-quality, potentially carefully
selected via greedy search, demonstrations in context, getting closer to the
performance of instruct models. Finally, we provide the first, to our
knowledge, systematic comparison of ICL and instruction fine-tuning (IFT) for
instruction following in the low data regime, where ICL can be a viable
alternative to IFT. Overall, our work advances the understanding of ICL as an
alignment technique and its relationship to IFT. We provide our code at
https://github.com/tml-epfl/icl-alignment.
| [
{
"version": "v1",
"created": "Thu, 30 May 2024 09:28:56 GMT"
},
{
"version": "v2",
"created": "Fri, 4 Oct 2024 12:39:20 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 12:31:18 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Zhao",
"Hao",
""
],
[
"Andriushchenko",
"Maksym",
""
],
[
"Croce",
"Francesco",
""
],
[
"Flammarion",
"Nicolas",
""
]
] |
2405.19876 | Alessio Mazzucchelli | Alessio Mazzucchelli, Adrian Garcia-Garcia, Elena Garces, Fernando
Rivas-Manzaneque, Francesc Moreno-Noguer, Adrian Penate-Sanchez | IReNe: Instant Recoloring of Neural Radiance Fields | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Advances in NERFs have allowed for 3D scene reconstructions and novel view
synthesis. Yet, efficiently editing these representations while retaining
photorealism is an emerging challenge. Recent methods face three primary
limitations: they're slow for interactive use, lack precision at object
boundaries, and struggle to ensure multi-view consistency. We introduce IReNe
to address these limitations, enabling swift, near real-time color editing in
NeRF. Leveraging a pre-trained NeRF model and a single training image with
user-applied color edits, IReNe swiftly adjusts network parameters in seconds.
This adjustment allows the model to generate new scene views, accurately
representing the color changes from the training image while also controlling
object boundaries and view-specific effects. Object boundary control is
achieved by integrating a trainable segmentation module into the model. The
process gains efficiency by retraining only the weights of the last network
layer. We observed that neurons in this layer can be classified into those
responsible for view-dependent appearance and those contributing to diffuse
appearance. We introduce an automated classification approach to identify these
neuron types and exclusively fine-tune the weights of the diffuse neurons. This
further accelerates training and ensures consistent color edits across
different views. A thorough validation on a new dataset, with edited object
colors, shows significant quantitative and qualitative advancements over
competitors, accelerating speeds by 5x to 500x.
| [
{
"version": "v1",
"created": "Thu, 30 May 2024 09:30:28 GMT"
},
{
"version": "v2",
"created": "Mon, 10 Jun 2024 12:12:38 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 11:39:12 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Mazzucchelli",
"Alessio",
""
],
[
"Garcia-Garcia",
"Adrian",
""
],
[
"Garces",
"Elena",
""
],
[
"Rivas-Manzaneque",
"Fernando",
""
],
[
"Moreno-Noguer",
"Francesc",
""
],
[
"Penate-Sanchez",
"Adrian",
""
]
] |
2405.20266 | Indranil Mukherjee | Indranil Mukherjee and P. K. Mohanty | Hyperuniformity in Ashkin-Teller model | 13 pages, 5 figs | J. Phys.: Condens. Matter 36 (2024) 465401 | 10.1088/1361-648X/ad6c99 | null | cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that equilibrium systems in $d$ dimension that obey the inequality
$d\nu> 2,$ known as Harris criterion, exhibit suppressed energy fluctuation in
their critical state. Ashkin-Teller model is an example in $d=2$ where the
correlation length exponent $\nu$ varies continuously with the inter-spin
interaction strength $\lambda$ and exceeds the value $\frac d2$ set by Harris
criterion when $\lambda$ is negative; there, the variance of the subsystem
energy across a length scale $l$ varies as $l^{d-\alpha}$ with hyperuniformity
exponent $\alpha = 2(1-\nu^{-1}).$ Point configurations constructed by
assigning unity to the sites which has coarse-grained energy beyond a threshold
value also exhibit suppressed number fluctuation and hyperuniformiyty with same
exponent $\alpha.$
| [
{
"version": "v1",
"created": "Thu, 30 May 2024 17:17:59 GMT"
},
{
"version": "v2",
"created": "Thu, 12 Sep 2024 16:47:19 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Mukherjee",
"Indranil",
""
],
[
"Mohanty",
"P. K.",
""
]
] |
2406.02360 | Sipan Aslan | Sipan Aslan and Hernando Ombao | Granger Causality in High-Dimensional Networks of Time Series | null | null | null | null | stat.AP stat.CO stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A novel approach is developed for discovering directed connectivity between
specified pairs of nodes in a high-dimensional network (HDN) of brain signals.
To accurately identify causal connectivity for such specified objectives, it is
necessary to properly address the influence of all other nodes within the
network. The proposed procedure herein starts with the estimation of a
low-dimensional representation of the other nodes in the network utilizing
(frequency-domain-based) spectral dynamic principal component analysis (sDPCA).
The resulting scores can then be removed from the nodes of interest, thus
eliminating the confounding effect of other nodes within the HDN. Accordingly,
causal interactions can be dissected between nodes that are isolated from the
effects of the network. Extensive simulations have demonstrated the
effectiveness of this approach as a tool for causality analysis in complex time
series networks. The proposed methodology has also been shown to be applicable
to multichannel EEG networks.
| [
{
"version": "v1",
"created": "Tue, 4 Jun 2024 14:38:23 GMT"
},
{
"version": "v2",
"created": "Sat, 5 Oct 2024 22:37:55 GMT"
},
{
"version": "v3",
"created": "Thu, 17 Apr 2025 20:21:05 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Aslan",
"Sipan",
""
],
[
"Ombao",
"Hernando",
""
]
] |
2406.03661 | Chandan Mazumdar | Shuvankar Gupta, Sudip Chakraborty, Vidha Bhasin, Celine Barreteau,
Jean-Claude Crivello, Jean-Marc Greneche, S.N. Jha, D. Bhattacharyya, Eric
Alleno, and Chandan Mazumdar | Randomness in atomic disorder and consequent squandering of
spin-polarization in a ferromagnetically fragile quaternary Heusler alloy
FeRuCrSi | null | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ru$_{2-x}$Fe$_x$CrSi ( 0 $<$ x $<$1) system is theoretically predicted to be
one of the very few known examples of robust half-metallic ferromagnet with
100\% spin polarization. Since Cr is considered to be the main contributor to
magnetism, the Fe/Ru substitution is not expected to disturb its magnetic
properties any significantly, and hence all Fe-containing members of the series
are predicted to follow Slater-Pauling rule with a saturation magnetic moment
of 2 ${\mu_B}$/f.u. However, contrarily to the theoretical expectations, some
experiments rather show a linear variation of the saturation magnetization and
Curie temperature with Fe (\textit{x}) substitution. The equiatomic member
FeRuCrSi of this family is also considered as a technologically important
material, where the band structure calculations suggest the material to be spin
gapless semiconductor. Through our in-depth structural analysis of FeRuCrSi
using X-ray diffraction, extended X-ray absorption fine structure and $^{57}$Fe
M\"{o}ssbauer spectrometry, we found a random disorder between Fe and Ru sites,
while the magnetic moment in this system is actually contributed by Fe atoms,
questioning the very basic foundation of the half-metallic character proposed
by all theoretical calculations on Ru$_{2-x}$Fe$_x$CrSi series. Our
M\"{o}ssbauer result also envisions a rather rare scenario where the main
physical properties are intricately correlated to the chemistry of the material
in the form of random atomic disorder on a localised scale.
| [
{
"version": "v1",
"created": "Thu, 6 Jun 2024 00:25:18 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 02:48:25 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Gupta",
"Shuvankar",
""
],
[
"Chakraborty",
"Sudip",
""
],
[
"Bhasin",
"Vidha",
""
],
[
"Barreteau",
"Celine",
""
],
[
"Crivello",
"Jean-Claude",
""
],
[
"Greneche",
"Jean-Marc",
""
],
[
"Jha",
"S. N.",
""
],
[
"Bhattacharyya",
"D.",
""
],
[
"Alleno",
"Eric",
""
],
[
"Mazumdar",
"Chandan",
""
]
] |
2406.04588 | Ting Tao | Ting Tao and Yitian Qian and Shaohua Pan | Convergence of the majorized PAM method with subspace correction for
low-rank composite factorization model | 34 pages | null | null | null | math.OC stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper focuses on the convergence certificates of the majorized proximal
alternating minimization (PAM) method with subspace correction, proposed in
\cite{TaoQianPan22} for the column $\ell_{2,0}$-norm regularized factorization
model and now extended to a class of low-rank composite factorization models
from matrix completion. The convergence analysis of this PAM method becomes
extremely challenging because a subspace correction step is introduced to every
proximal subproblem to ensure a closed-form solution. We establish the full
convergence of the iterate sequence and column subspace sequences of factor
pairs generated by the PAM, under the KL property of the objective function and
a condition that holds automatically for the column $\ell_{2,0}$-norm function.
Numerical comparison with the popular proximal alternating linearized
minimization (PALM) method is conducted on one-bit matrix completion problems,
which indicates that the PAM with subspace correction has an advantage in
seeking lower relative error within less time.
| [
{
"version": "v1",
"created": "Fri, 7 Jun 2024 02:33:22 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 09:23:32 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Tao",
"Ting",
""
],
[
"Qian",
"Yitian",
""
],
[
"Pan",
"Shaohua",
""
]
] |
2406.06462 | Suyuchen Wang | Tianyu Zhang, Suyuchen Wang, Lu Li, Ge Zhang, Perouz Taslakian, Sai
Rajeswar, Jie Fu, Bang Liu, Yoshua Bengio | VCR: A Task for Pixel-Level Complex Reasoning in Vision Language Models
via Restoring Occluded Text | Accepted at ICLR 2025. Original paper name: VCR: Visual Caption
Restoration | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | We introduce Visual Caption Restoration (VCR), a novel vision-language task
that challenges models to accurately restore partially obscured texts using
pixel-level hints within images. This task stems from the observation that text
embedded in images is intrinsically different from common visual elements and
natural language due to the need to align the modalities of vision, text, and
text embedded in images. While numerous works have integrated text embedded in
images into visual question-answering tasks, approaches to these tasks
generally rely on optical character recognition or masked language modeling,
thus reducing the task to mainly text-based processing. However, text-based
processing becomes ineffective in VCR as accurate text restoration depends on
the combined information from provided images, context, and subtle cues from
the tiny exposed areas of masked texts. We develop a pipeline to generate
synthetic images for the VCR task using image-caption pairs, with adjustable
caption visibility to control the task difficulty. With this pipeline, we
construct a dataset for VCR called VCR-Wiki using images with captions from
Wikipedia, comprising 2.11M English and 346K Chinese entities in both easy and
hard split variants. Our results reveal that current vision language models
significantly lag behind human performance in the VCR task, and merely
fine-tuning the models on our dataset does not lead to notable improvements. We
release VCR-Wiki and the data construction code to facilitate future research.
| [
{
"version": "v1",
"created": "Mon, 10 Jun 2024 16:58:48 GMT"
},
{
"version": "v2",
"created": "Mon, 24 Jun 2024 07:05:01 GMT"
},
{
"version": "v3",
"created": "Fri, 4 Oct 2024 08:22:48 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 16:42:48 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Zhang",
"Tianyu",
""
],
[
"Wang",
"Suyuchen",
""
],
[
"Li",
"Lu",
""
],
[
"Zhang",
"Ge",
""
],
[
"Taslakian",
"Perouz",
""
],
[
"Rajeswar",
"Sai",
""
],
[
"Fu",
"Jie",
""
],
[
"Liu",
"Bang",
""
],
[
"Bengio",
"Yoshua",
""
]
] |
2406.07172 | Vyacheslav Storchak G. | Oleg E. Parfenov, Dmitry V. Averyanov, Ivan S. Sokolov, Alexey N.
Mihalyuk, Oleg A. Kondratev, Alexander N. Taldenkov, Andrey M. Tokmachev,
Vyacheslav G. Storchak | Pushing an Altermagnet to the Ultimate 2D Limit: Evidence of Symmetry
Breaking in Monolayers of GdAlSi | null | JACS 147, 5911 (2025) | 10.1021/jacs.4c14891 | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Altermagnets have emerged as a class of materials combining certain
ferromagnetic properties with zero net magnetization. This combination is
highly promising for spintronics, especially if a material can be brought to a
nanoscale size. However, experimental studies of the 2D limit of altermagnets
and evolution of their properties with thickness are lacking. Here, we study
epitaxial films on silicon of the Weyl altermagnet GdAlSi ranging from more
than a hundred unit cells to a single unit cell. The films are synthesized by
molecular beam epitaxy and, expectedly, do not show any discernible net
magnetic moments. Electron transport studies reveal a remarkable transformation
of the electron state with the film thickness. Thick films exhibit negative
longitudinal magnetoresistance associated with the chiral anomaly but do not
demonstrate altermagnetic properties in electron transport due to symmetry
restrictions. In ultrathin films, a spontaneous anomalous Hall effect manifests
itself, indicating a non-relativistic spin splitting in the electronic
structure. The transformation is associated with crystal symmetry breaking
accompanying the 3D-to-2D crossover. The work highlights the role of
dimensionality in altermagnetism and provides a platform for studies of
altermagnets aiming at ultra-compact spintronics.
| [
{
"version": "v1",
"created": "Tue, 11 Jun 2024 11:28:53 GMT"
},
{
"version": "v2",
"created": "Tue, 4 Feb 2025 09:41:08 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Parfenov",
"Oleg E.",
""
],
[
"Averyanov",
"Dmitry V.",
""
],
[
"Sokolov",
"Ivan S.",
""
],
[
"Mihalyuk",
"Alexey N.",
""
],
[
"Kondratev",
"Oleg A.",
""
],
[
"Taldenkov",
"Alexander N.",
""
],
[
"Tokmachev",
"Andrey M.",
""
],
[
"Storchak",
"Vyacheslav G.",
""
]
] |
2406.07226 | Angela Rosy Morgillo | Angela Rosy Morgillo, Massimiliano F. Sacchi and Chiara Macchiavello | Detecting Markovianity of Quantum Processes via Recurrent Neural
Networks | This version includes results obtained with a noisy version of the
dataset, along with other minor improvements | Quantum Mach. Intell. 7, 47 (2025) | 10.1007/s42484-025-00252-w | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We present a novel methodology utilizing Recurrent Neural Networks (RNNs) to
classify Markovian and non-Markovian quantum processes, leveraging time series
data derived from Choi states. The model exhibits exceptional accuracy,
surpassing 95%, across diverse scenarios, including dephasing and Pauli
channels in an arbitrary basis, generalized amplitude damping dynamics, and
even in the presence of noise. Additionally, the developed model shows
efficient forecasting capabilities for the analyzed time series data. These
results suggest the potential of RNNs in discerning and predicting the
Markovian nature of quantum processes.
| [
{
"version": "v1",
"created": "Tue, 11 Jun 2024 13:05:36 GMT"
},
{
"version": "v2",
"created": "Wed, 12 Jun 2024 14:36:52 GMT"
},
{
"version": "v3",
"created": "Thu, 27 Mar 2025 19:14:11 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Morgillo",
"Angela Rosy",
""
],
[
"Sacchi",
"Massimiliano F.",
""
],
[
"Macchiavello",
"Chiara",
""
]
] |
2406.11165 | Zhen-Guo Fu | Yuan Hong, Zhen-Guo Fu, Zhou-Wei-Yu Chen, Feng Chi, Zhigang Wang, Wei
Zhang and Ping Zhang | Tunable Fano and Dicke effects in quantum transport of double quantum
dots sandwiched between topological insulators | 7 pages, 6 figures | null | 10.1103/PhysRevB.111.155423 | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | We study the quantum transport in double quantum dots (DQD) sandwiched
between surfaces of topological insulator (TI) Bi$_{2}$Te$_{3}$, which possess
strong spin-orbit coupling (SOC) and $^{d}$C$_{3v}$ double group symmetry.
Different from the spin-conserved case with two-dimensional electron gas (2DEG)
electrodes, the conductance displays a universal scaling relation for different
Fermi energy associated with the topological nature/linear dispersion of
topological surface states. The interplay between direct inter-dot tunneling
and surface state mediated interaction leads to tunable Dicke and Fano effects
with changing the inter-dot distance. We propose nano-rulers with different
measurement range and resolution based on the Fano $q$-factor. Furthermore,
when applying an in-plane Zeeman field, a crossover from a double-peak shape to
a quad-peak shape in conductance curve appears. Moreover, the rotational
symmetry of the system could also be revealed from the conductance pattern. Our
findings contribute to a better understanding of the quantum transport in the
presence of electrode's SOC topological states.
| [
{
"version": "v1",
"created": "Mon, 17 Jun 2024 03:09:24 GMT"
},
{
"version": "v2",
"created": "Wed, 26 Mar 2025 16:01:07 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Hong",
"Yuan",
""
],
[
"Fu",
"Zhen-Guo",
""
],
[
"Chen",
"Zhou-Wei-Yu",
""
],
[
"Chi",
"Feng",
""
],
[
"Wang",
"Zhigang",
""
],
[
"Zhang",
"Wei",
""
],
[
"Zhang",
"Ping",
""
]
] |
2406.11260 | Sungwon Park | Sungwon Park, Sungwon Han, Xing Xie, Jae-Gil Lee, Meeyoung Cha | Adversarial Style Augmentation via Large Language Model for Robust Fake
News Detection | WWW'25 research track accepted | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The spread of fake news harms individuals and presents a critical social
challenge that must be addressed. Although numerous algorithmic and insightful
features have been developed to detect fake news, many of these features can be
manipulated with style-conversion attacks, especially with the emergence of
advanced language models, making it more difficult to differentiate from
genuine news. This study proposes adversarial style augmentation, AdStyle,
designed to train a fake news detector that remains robust against various
style-conversion attacks. The primary mechanism involves the strategic use of
LLMs to automatically generate a diverse and coherent array of style-conversion
attack prompts, enhancing the generation of particularly challenging prompts
for the detector. Experiments indicate that our augmentation strategy
significantly improves robustness and detection performance when evaluated on
fake news benchmark datasets.
| [
{
"version": "v1",
"created": "Mon, 17 Jun 2024 07:00:41 GMT"
},
{
"version": "v2",
"created": "Mon, 22 Jul 2024 11:56:44 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 07:15:33 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Park",
"Sungwon",
""
],
[
"Han",
"Sungwon",
""
],
[
"Xie",
"Xing",
""
],
[
"Lee",
"Jae-Gil",
""
],
[
"Cha",
"Meeyoung",
""
]
] |
2406.12307 | ChaeHun Park | Seungbin Yang, ChaeHun Park, Taehee Kim, Jaegul Choo | Can Tool-augmented Large Language Models be Aware of Incomplete
Conditions? | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Recent advancements in integrating large language models (LLMs) with tools
have allowed the models to interact with real-world environments. However,
these tool-augmented LLMs often encounter incomplete scenarios when users
provide partial information or the necessary tools are unavailable. Recognizing
and managing such scenarios is crucial for LLMs to ensure their reliability,
but this exploration remains understudied. This study examines whether LLMs can
identify incomplete conditions and appropriately determine when to refrain from
using tools. To this end, we address a dataset by manipulating instances from
two datasets by removing necessary tools or essential information for tool
invocation. Our experiments show that LLMs often struggle to identify the
absence of information required to utilize specific tools and recognize the
absence of appropriate tools. We further analyze model behaviors in different
environments and compare their performance against humans. Our research can
contribute to advancing reliable LLMs by addressing common scenarios during
interactions between humans and LLMs. Our code and dataset will be publicly
available.
| [
{
"version": "v1",
"created": "Tue, 18 Jun 2024 06:28:06 GMT"
},
{
"version": "v2",
"created": "Sun, 29 Sep 2024 05:11:45 GMT"
},
{
"version": "v3",
"created": "Tue, 17 Dec 2024 05:33:04 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 13:07:08 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Yang",
"Seungbin",
""
],
[
"Park",
"ChaeHun",
""
],
[
"Kim",
"Taehee",
""
],
[
"Choo",
"Jaegul",
""
]
] |
2406.12319 | ChaeHun Park | Hawon Jeong, ChaeHun Park, Jimin Hong, Hojoon Lee, Jaegul Choo | The Comparative Trap: Pairwise Comparisons Amplifies Biased Preferences
of LLM Evaluators | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | As large language models (LLMs) are increasingly used as evaluators for
natural language generation tasks, ensuring unbiased assessments is essential.
However, LLM evaluators often display biased preferences, such as favoring
verbosity and authoritative tones. Our empirical analysis reveals that these
biases are exacerbated in pairwise evaluation, where LLMs directly compare two
outputs and easily prioritize superficial attributes. In contrast, pointwise
evaluation, which assesses outputs independently, is less susceptible to such
bias because each output is judged in isolation. To address the limitations of
the pairwise evaluation, we introduce a novel evaluation method, PRePair, which
integrates pointwise reasoning within a pairwise framework. PRePair effectively
alleviates biased preference, improving performance on the adversarial
benchmark (LLMBar) while outperforming pointwise evaluation on the standard
benchmark (MT-Bench).
| [
{
"version": "v1",
"created": "Tue, 18 Jun 2024 06:43:04 GMT"
},
{
"version": "v2",
"created": "Thu, 3 Oct 2024 09:38:48 GMT"
},
{
"version": "v3",
"created": "Wed, 16 Oct 2024 13:39:49 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 13:12:42 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Jeong",
"Hawon",
""
],
[
"Park",
"ChaeHun",
""
],
[
"Hong",
"Jimin",
""
],
[
"Lee",
"Hojoon",
""
],
[
"Choo",
"Jaegul",
""
]
] |
2406.13211 | Keshav V | Keshav V, M.S. Santhanam | Amplitude Amplification and Estimation using a Floquet system | 10 pages, 9 figures | Phys. Rev. A 111, 032601 (2025) | 10.1103/PhysRevA.111.032601 | null | quant-ph cond-mat.quant-gas | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The quantum kicked rotor (QKR) is a fundamental model of time-dependent
quantum chaos and the physics of Anderson localization. It is one of the most
well-studied Floquet systems. In this work, it is shown that QKR can be used to
implement a quantum algorithm to perform unstructured search; namely Amplitude
Amplification, a generalization of Grover's search algorithm. Further, the QKR
is employed for amplitude estimation when the amplitude of the marked states is
unknown. It is also shown that the characteristic property of dynamical
localization of the QKR can be exploited to enhance the performance of the
amplitude amplification algorithm by reducing its average runtime. The
sensitivity of the success probability of unstructured search to detuning from
resonance and the effects of noisy kick strengths are analyzed and the
robustness of the QKR based algorithm is demonstrated. The experimental
feasibility of every component of the algorithm is discussed.
| [
{
"version": "v1",
"created": "Wed, 19 Jun 2024 04:44:12 GMT"
},
{
"version": "v2",
"created": "Thu, 29 Aug 2024 05:14:56 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"V",
"Keshav",
""
],
[
"Santhanam",
"M. S.",
""
]
] |
2406.15449 | Simone Pigolotti | Samuel Cure, Florian G. Pflug, Simone Pigolotti | Exponential rate of epidemic spreading on complex networks | 15 pages, 13 figures, accepted version | Phys. Rev. E 111, 044311 (2025) | 10.1103/PhysRevE.111.044311 | null | physics.soc-ph cond-mat.stat-mech q-bio.PE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The initial phase of an epidemic is often characterized by an exponential
increase in the number of infected individuals. In this paper, we predict the
exponential spreading rate of an epidemic on a complex network. We first find
an expression of the reproduction number for a network, based on the degree
distribution, the network assortativity, and the level of clustering. We then
connect this reproduction number and the disease infectiousness to the
spreading rate. Our result holds for a broad range of networks, apart from
networks with very broad degree distribution, where no clear exponential regime
is present. Our theory bridges the gap between classic epidemiology and the
theory of complex networks, with broad implications for model inference and
policy making.
| [
{
"version": "v1",
"created": "Wed, 5 Jun 2024 08:11:41 GMT"
},
{
"version": "v2",
"created": "Tue, 25 Jun 2024 05:26:57 GMT"
},
{
"version": "v3",
"created": "Mon, 5 Aug 2024 01:41:55 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 07:31:11 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Cure",
"Samuel",
""
],
[
"Pflug",
"Florian G.",
""
],
[
"Pigolotti",
"Simone",
""
]
] |
2406.16039 | Oluwatosin Alabi | Oluwatosin Alabi, Ko Ko Zayar Toe, Zijian Zhou, Charlie Budd, Nicholas
Raison, Miaojing Shi, Tom Vercauteren | CholecInstanceSeg: A Tool Instance Segmentation Dataset for Laparoscopic
Surgery | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In laparoscopic and robotic surgery, precise tool instance segmentation is an
essential technology for advanced computer-assisted interventions. Although
publicly available procedures of routine surgeries exist, they often lack
comprehensive annotations for tool instance segmentation. Additionally, the
majority of standard datasets for tool segmentation are derived from
porcine(pig) surgeries. To address this gap, we introduce CholecInstanceSeg,
the largest open-access tool instance segmentation dataset to date. Derived
from the existing CholecT50 and Cholec80 datasets, CholecInstanceSeg provides
novel annotations for laparoscopic cholecystectomy procedures in patients. Our
dataset comprises 41.9k annotated frames extracted from 85 clinical procedures
and 64.4k tool instances, each labelled with semantic masks and instance IDs.
To ensure the reliability of our annotations, we perform extensive quality
control, conduct label agreement statistics, and benchmark the segmentation
results with various instance segmentation baselines. CholecInstanceSeg aims to
advance the field by offering a comprehensive and high-quality open-access
dataset for the development and evaluation of tool instance segmentation
algorithms.
| [
{
"version": "v1",
"created": "Sun, 23 Jun 2024 07:30:06 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 09:41:09 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Alabi",
"Oluwatosin",
""
],
[
"Toe",
"Ko Ko Zayar",
""
],
[
"Zhou",
"Zijian",
""
],
[
"Budd",
"Charlie",
""
],
[
"Raison",
"Nicholas",
""
],
[
"Shi",
"Miaojing",
""
],
[
"Vercauteren",
"Tom",
""
]
] |
2406.16733 | Luca Sabatini | Daniele Dona, Luca Sabatini | The diameter of random Schreier graphs | 10 pages, to appear in European J. Combin | null | null | null | math.CO math.GR | http://creativecommons.org/licenses/by/4.0/ | We give a combinatorial proof of the following theorem. Let $G$ be any finite
group acting transitively on a set of cardinality $n$. If $S \subseteq G$ is a
random set of size $k$, with $k \geq (\log n)^{1+\varepsilon}$ for some
$\varepsilon >0$, then the diameter of the corresponding Schreier graph is
$O(\log_k n)$ with high probability. Except for the implicit constant, this
result is the best possible.
| [
{
"version": "v1",
"created": "Mon, 24 Jun 2024 15:39:03 GMT"
},
{
"version": "v2",
"created": "Mon, 1 Jul 2024 02:49:11 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 10:11:59 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Dona",
"Daniele",
""
],
[
"Sabatini",
"Luca",
""
]
] |
2406.17000 | Tianyang Liu | Tianyang Liu, Jiajun Zhang, Yuan Shi, Junhua Gu, Quan Guo, Yidong Xu,
Furen Deng, Fengquan Wu, Yanping Cong and Xuelei Chen | Forecast measurement of the 21 cm global spectrum from Lunar orbit with
the Vari-Zeroth-Order Polynomial (VZOP) method | 20 pages, 19 figures, 2 tables, Accepted for publication in PASA | null | 10.1017/pasa.2025.17 | null | astro-ph.IM astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The cosmic 21 cm signal serves as a crucial probe for studying the
evolutionary history of the Universe. However, detecting the 21 cm signal poses
significant challenges due to its extremely faint nature. To mitigate the
interference from the Earth's radio frequency interference (RFI), the ground
and the ionospheric effects, the Discovering the Sky at the Longest Wavelength
(DSL) project will deploy a constellation of satellites in Lunar orbit, with
its high-frequency daughter satellite tasked with detecting the global 21 cm
signal from cosmic dawn and reionization era (CD/EoR). We intend to employ the
Vari-Zeroth-Order Polynomial (VZOP) for foreground fitting and subtracting. We
have studied the effect of thermal noise, thermal radiation from the Moon, the
Lunar reflection, anisotropic frequency-dependent beam, inaccurate antenna beam
pattern, and RFI contamination. We discovered that the RFI contamination can
significantly affect the fitting process and thus prevent us from detecting the
signal. Therefore, experimenting on the far side of the moon is crucial. We
also discovered that using VZOP together with DSL, after 1080 orbits around the
Moon, which takes about 103 days, we can successfully detect the CD/EoR 21 cm
signal.
| [
{
"version": "v1",
"created": "Mon, 24 Jun 2024 12:43:51 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 09:10:25 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Liu",
"Tianyang",
""
],
[
"Zhang",
"Jiajun",
""
],
[
"Shi",
"Yuan",
""
],
[
"Gu",
"Junhua",
""
],
[
"Guo",
"Quan",
""
],
[
"Xu",
"Yidong",
""
],
[
"Deng",
"Furen",
""
],
[
"Wu",
"Fengquan",
""
],
[
"Cong",
"Yanping",
""
],
[
"Chen",
"Xuelei",
""
]
] |
2406.17767 | Sebastian Perez-Salazar | Johannes Brustle, Sebastian Perez-Salazar, Victor Verdugo | Splitting Guarantees for Prophet Inequalities via Nonlinear Systems | null | null | null | null | cs.GT math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The prophet inequality is one of the cornerstone problems in optimal stopping
theory and has become a crucial tool for designing sequential algorithms in
Bayesian settings. In the i.i.d. $k$-selection prophet inequality problem, we
sequentially observe $n$ non-negative random values sampled from a known
distribution. Each time, a decision is made to accept or reject the value, and
under the constraint of accepting at most $k$. For $k=1$, Hill and Kertz [Ann.
Probab. 1982] provided an upper bound on the worst-case approximation ratio
that was later matched by an algorithm of Correa et al. [Math. Oper. Res.
2021]. The worst-case tight approximation ratio for $k=1$ is computed by
studying a differential equation that naturally appears when analyzing the
optimal dynamic programming policy. A similar result for $k>1$ has remained
elusive.
In this work, we introduce a nonlinear system of differential equations for
the i.i.d. $k$-selection prophet inequality that generalizes Hill and Kertz's
equation when $k=1$. Our nonlinear system is defined by $k$ constants that
determine its functional structure, and their summation provides a lower bound
on the optimal policy's asymptotic approximation ratio for the i.i.d.
$k$-selection prophet inequality. To obtain this result, we introduce for every
$k$ an infinite-dimensional linear programming formulation that fully
characterizes the worst-case tight approximation ratio of the $k$-selection
prophet inequality problem for every $n$, and then we follow a dual-fitting
approach to link with our nonlinear system for sufficiently large values of
$n$. As a corollary, we use our provable lower bounds to establish a tight
approximation ratio for the stochastic sequential assignment problem in the
i.i.d. non-negative regime.
| [
{
"version": "v1",
"created": "Tue, 25 Jun 2024 17:49:59 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Apr 2025 22:27:15 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Brustle",
"Johannes",
""
],
[
"Perez-Salazar",
"Sebastian",
""
],
[
"Verdugo",
"Victor",
""
]
] |
2406.19511 | Mathilde Gerbelli-Gauthier | Mathilde Gerbelli-Gauthier, Akshay Venkatesh | Cohomology of Fuchsian groups and Fourier interpolation | 29 pages. Minor edits and new example in Section 6. Comments Welcome! | null | null | null | math.NT math.FA math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We give a new proof of a Fourier interpolation result first proved by
Radchenko-Viazovska, deriving it from a vanishing result of the first
cohomology of a Fuchsian group with coefficients in the Weil representation.
| [
{
"version": "v1",
"created": "Thu, 27 Jun 2024 20:16:24 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Apr 2025 20:22:55 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Gerbelli-Gauthier",
"Mathilde",
""
],
[
"Venkatesh",
"Akshay",
""
]
] |
2406.20091 | Takato Yoshimura | Takato Yoshimura and \v{Z}iga Krajnik | Anomalous current fluctuations from Euler hydrodynamics | v1:6 + 17 pages, 2 + 2 figures, v2:minor changes, additional
references, v3: final version, results on the diffusion constant, which will
be reported elsewhere, dropped from supplemental materials (now incorporated
in the main text as Appendices), the typos on the location of the absolute
value in Eqs. (C4, C5) fixed (these typos were unnoticed in the published
version) | Phys. Rev. E 111, 024141 (2025) | 10.1103/PhysRevE.111.024141 | null | cond-mat.stat-mech cond-mat.quant-gas cond-mat.str-el math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We consider the hydrodynamic origin of anomalous current fluctuations in a
family of stochastic charged cellular automata. Using ballistic macroscopic
fluctuation theory, we study both typical and large fluctuations of the charge
current and reproduce microscopic results which are available for the
deterministic single-file limit of the models. Our results indicate that in
general initial fluctuations propagated by Euler equations fully characterize
both scales of anomalous fluctuations. For stochastic dynamics, we find an
additional contribution to typical fluctuations and conjecture the functional
form of the typical probability distribution, which we confirm by numerical
simulations.
| [
{
"version": "v1",
"created": "Fri, 28 Jun 2024 17:57:05 GMT"
},
{
"version": "v2",
"created": "Fri, 12 Jul 2024 11:48:26 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 11:50:24 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Yoshimura",
"Takato",
""
],
[
"Krajnik",
"Žiga",
""
]
] |
2407.00641 | Rachmad Vidya Wicaksana Putra | Rachmad Vidya Wicaksana Putra, Muhammad Shafique | NeuroNAS: Enhancing Efficiency of Neuromorphic In-Memory Computing for
Intelligent Mobile Agents through Hardware-Aware Spiking Neural Architecture
Search | 9 pages, 14 figures, 2 tables | null | null | null | cs.NE cs.AI cs.AR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Intelligent mobile agents (e.g., UGVs and UAVs) typically demand low
power/energy consumption when solving their machine learning (ML)-based tasks,
since they are usually powered by portable batteries with limited capacity. A
potential solution is employing neuromorphic computing with Spiking Neural
Networks (SNNs), which leverages event-based computation to enable ultra-low
power/energy ML algorithms. To maximize the performance efficiency of SNN
inference, the In-Memory Computing (IMC)-based hardware accelerators with
emerging device technologies (e.g., RRAM) can be employed. However, SNN models
are typically developed without considering constraints from the application
and the underlying IMC hardware, thereby hindering SNNs from reaching their
full potential in performance and efficiency. To address this, we propose
NeuroNAS, a novel framework for developing energyefficient neuromorphic IMC for
intelligent mobile agents using hardware-aware spiking neural architecture
search (NAS), i.e., by quickly finding an SNN architecture that offers high
accuracy under the given constraints (e.g., memory, area, latency, and energy
consumption). Its key steps include: optimizing SNN operations to enable
efficient NAS, employing quantization to minimize the memory footprint,
developing an SNN architecture that facilitates an effective learning, and
devising a systematic hardware-aware search algorithm to meet the constraints.
Compared to the state-of-the-art techniques, NeuroNAS quickly finds SNN
architectures (with 8bit weight precision) that maintain high accuracy by up to
6.6x search time speed-ups, while achieving up to 92% area savings, 1.2x
latency improvements, 84% energy savings across different datasets (i.e.,
CIFAR-10, CIFAR-100, and TinyImageNet-200); while the state-of-the-art fail to
meet all constraints at once.
| [
{
"version": "v1",
"created": "Sun, 30 Jun 2024 09:51:58 GMT"
},
{
"version": "v2",
"created": "Fri, 6 Dec 2024 08:35:27 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 07:00:39 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Putra",
"Rachmad Vidya Wicaksana",
""
],
[
"Shafique",
"Muhammad",
""
]
] |
2407.00813 | Qi Deng | Qi Deng | Liquidity Adjustment in Multivariate Volatility Modeling: Evidence from
Portfolios of Cryptocurrencies and US Stocks | 48 pages, 1 figure | null | null | null | q-fin.ST q-fin.PM | http://creativecommons.org/licenses/by/4.0/ | We develop a liquidity-sensitive multivariate volatility framework to improve
the estimation of time-varying covariance structures under market frictions. We
introduce two novel portfolio-level liquidity measures, liquidity jump and
liquidity diffusion, which capture magnitude and volatility of liquidity
fluctuation, respectively, and construct liquidity-adjusted return and
volatility that reflect real-time liquidity variability. These
liquidity-adjusted inputs are integrated into a VECM-DCC/ADCC-Bayesian model,
allowing for conditional and posterior covariance estimation under liquidity
stress. Applying this framework to portfolios of cryptocurrencies and US
stocks, we find that traditional models misrepresent volatility and
co-movement, while liquidity-adjusted models yield more stable and
interpretable risk structures, particularly for portfolios of cryptocurrencies.
The findings support the use of liquidity-adjusted multivariate models as
statistically grounded tools for assessing the propagation of portfolio risk
under market frictions, with implications for asset pricing, market
microstructure design, and portfolio management.
| [
{
"version": "v1",
"created": "Sat, 30 Mar 2024 00:03:21 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Apr 2025 21:47:04 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Deng",
"Qi",
""
]
] |
2407.01590 | Angelo Rosa Prof. Dr. | Davide Marcato and Achille Giacometti and Amos Maritan and Angelo Rosa | Theory of polymers in binary solvent solutions: mean-field free energy
and phase behavior | 12 pages, 6 figures (main); 9 pages, 5 figures (suppl info); Physical
Review Materials, accepted for publication | Physical Review Materials 8, 125601 (2024) | 10.1103/PhysRevMaterials.8.125601 | null | cond-mat.soft | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a lattice model for polymer solutions, explicitly incorporating
interactions with a bath of solvent and cosolvent molecules. By exploiting the
well-known analogy between polymer systems and the $O(n)$-vector spin model in
the limit $n \to 0$, we derive an exact field-theoretic expression for the
partition function of the system. The latter is then evaluated at the saddle
point, providing a mean-field estimate of the free energy. The resulting
expression, which conforms to the Flory-Huggins type, is then used to analyze
the system's stability with respect to phase separation, complemented by a
numerical approach based on convex hull evaluation. We demonstrate that this
simple lattice model can effectively explain the behavior of a variety of
seemingly unrelated polymer systems, which have been predominantly investigated
in the past only through numerical simulations. This includes both,
single-chain and multi-chain, solutions. Our findings emphasize the
fundamental, mutually competing, roles of solvent and cosolvent in polymer
systems.
| [
{
"version": "v1",
"created": "Sat, 22 Jun 2024 15:59:47 GMT"
},
{
"version": "v2",
"created": "Wed, 18 Dec 2024 17:01:40 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Marcato",
"Davide",
""
],
[
"Giacometti",
"Achille",
""
],
[
"Maritan",
"Amos",
""
],
[
"Rosa",
"Angelo",
""
]
] |
2407.02694 | Daniel Jeong | Daniel P. Jeong, Zachary C. Lipton, Pradeep Ravikumar | LLM-Select: Feature Selection with Large Language Models | Published in Transactions on Machine Learning Research (TMLR), April
2025 | null | null | null | cs.LG cs.AI cs.CL stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we demonstrate a surprising capability of large language
models (LLMs): given only input feature names and a description of a prediction
task, they are capable of selecting the most predictive features, with
performance rivaling the standard tools of data science. Remarkably, these
models exhibit this capacity across various query mechanisms. For example, we
zero-shot prompt an LLM to output a numerical importance score for a feature
(e.g., "blood pressure") in predicting an outcome of interest (e.g., "heart
failure"), with no additional context. In particular, we find that the latest
models, such as GPT-4, can consistently identify the most predictive features
regardless of the query mechanism and across various prompting strategies. We
illustrate these findings through extensive experiments on real-world data,
where we show that LLM-based feature selection consistently achieves strong
performance competitive with data-driven methods such as the LASSO, despite
never having looked at the downstream training data. Our findings suggest that
LLMs may be useful not only for selecting the best features for training but
also for deciding which features to collect in the first place. This could
benefit practitioners in domains like healthcare and the social sciences, where
collecting high-quality data comes at a high cost.
| [
{
"version": "v1",
"created": "Tue, 2 Jul 2024 22:23:40 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Apr 2025 21:50:37 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Jeong",
"Daniel P.",
""
],
[
"Lipton",
"Zachary C.",
""
],
[
"Ravikumar",
"Pradeep",
""
]
] |
2407.04129 | Wai-Keong Mok | Wai-Keong Mok, Stuart J. Masson, Dan M. Stamper-Kurn, Tanya
Zelevinsky, Ana Asenjo-Garcia | Ground-state selection via many-body superradiant decay | 5 pages, 3 figures | Phys. Rev. Research 7, L022015 (2025) | 10.1103/PhysRevResearch.7.L022015 | null | quant-ph physics.atm-clus physics.atom-ph physics.optics | http://creativecommons.org/licenses/by/4.0/ | For a single particle, relaxation into different ground states is governed by
fixed branching ratios determined by the transition matrix element and the
environment. Here, we show that in many-body open quantum systems the
occupation probability of one ground state can be boosted well beyond what is
dictated by single-particle branching ratios. Despite the competition,
interactions suppress all but the dominant decay transition, leading to a
'winner takes all' dynamic where the system primarily settles into the dominant
ground state. We prove that, in the presence of permutation symmetry, this
problem is exactly solvable for any number of competing channels. Additionally,
we develop an approximate model for the dynamics by mapping the evolution onto
a fluid continuity equation, and analytically demonstrate that the dominant
transition ratio converges to unity as a power law with increasing system size,
for any branching ratios. This near-deterministic preparation of the dominant
ground state has broad applicability. As an example, we discuss a protocol for
molecular photoassociation where collective dynamics effectively acts as a
catalyst, amplifying the yield in a specific final state. Our results open new
avenues for many-body strategies in the preparation and control of quantum
systems.
| [
{
"version": "v1",
"created": "Thu, 4 Jul 2024 19:13:48 GMT"
},
{
"version": "v2",
"created": "Thu, 13 Mar 2025 18:37:31 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Mok",
"Wai-Keong",
""
],
[
"Masson",
"Stuart J.",
""
],
[
"Stamper-Kurn",
"Dan M.",
""
],
[
"Zelevinsky",
"Tanya",
""
],
[
"Asenjo-Garcia",
"Ana",
""
]
] |
2407.05441 | Leheng Sheng | Leheng Sheng and An Zhang and Yi Zhang and Yuxin Chen and Xiang Wang
and Tat-Seng Chua | Language Representations Can be What Recommenders Need: Findings and
Potentials | ICLR 2025 (Oral). Codes are available at
https://github.com/LehengTHU/AlphaRec | null | null | null | cs.IR cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent studies empirically indicate that language models (LMs) encode rich
world knowledge beyond mere semantics, attracting significant attention across
various fields. However, in the recommendation domain, it remains uncertain
whether LMs implicitly encode user preference information. Contrary to
prevailing understanding that LMs and traditional recommenders learn two
distinct representation spaces due to the huge gap in language and behavior
modeling objectives, this work re-examines such understanding and explores
extracting a recommendation space directly from the language representation
space. Surprisingly, our findings demonstrate that item representations, when
linearly mapped from advanced LM representations, yield superior recommendation
performance. This outcome suggests the possible homomorphism between the
advanced language representation space and an effective item representation
space for recommendation, implying that collaborative signals may be implicitly
encoded within LMs. Motivated by these findings, we explore the possibility of
designing advanced collaborative filtering (CF) models purely based on language
representations without ID-based embeddings. To be specific, we incorporate
several crucial components to build a simple yet effective model, with item
titles as the input. Empirical results show that such a simple model can
outperform leading ID-based CF models, which sheds light on using language
representations for better recommendation. Moreover, we systematically analyze
this simple model and find several key features for using advanced language
representations: a good initialization for item representations, zero-shot
recommendation abilities, and being aware of user intention. Our findings
highlight the connection between language modeling and behavior modeling, which
can inspire both natural language processing and recommender system
communities.
| [
{
"version": "v1",
"created": "Sun, 7 Jul 2024 17:05:24 GMT"
},
{
"version": "v2",
"created": "Thu, 3 Oct 2024 03:41:56 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 05:54:01 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Sheng",
"Leheng",
""
],
[
"Zhang",
"An",
""
],
[
"Zhang",
"Yi",
""
],
[
"Chen",
"Yuxin",
""
],
[
"Wang",
"Xiang",
""
],
[
"Chua",
"Tat-Seng",
""
]
] |
2407.06965 | Jaeseong Oh | Jim Haglund, Jaeseong Oh, Meesue Yoo | $\alpha$-chromatic symmetric functions | Minor corrections | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we introduce the \emph{$\alpha$-chromatic symmetric functions}
$\chi^{(\alpha)}_\pi[X;q]$, extending Shareshian and Wachs' chromatic symmetric
functions with an additional real parameter $\alpha$. We present positive
combinatorial formulas with explicit interpretations. Notably, we show an
explicit monomial expansion in terms of the $\alpha$-binomial basis and an
expansion into certain chromatic symmetric functions in terms of the
$\alpha$-falling factorial basis. Among various connections with other
subjects, we highlight a significant link to $q$-rook theory, including a new
solution to the $q$-hit problem posed by Garsia and Remmel in their 1986 paper
introducing $q$-rook polynomials.
| [
{
"version": "v1",
"created": "Tue, 9 Jul 2024 15:45:19 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 01:59:30 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Haglund",
"Jim",
""
],
[
"Oh",
"Jaeseong",
""
],
[
"Yoo",
"Meesue",
""
]
] |
2407.07880 | Junkang Wu | Junkang Wu, Yuexiang Xie, Zhengyi Yang, Jiancan Wu, Jiawei Chen,
Jinyang Gao, Bolin Ding, Xiang Wang, Xiangnan He | Towards Robust Alignment of Language Models: Distributionally
Robustifying Direct Preference Optimization | null | null | null | null | cs.LG cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study addresses the challenge of noise in training datasets for Direct
Preference Optimization (DPO), a method for aligning Large Language Models
(LLMs) with human preferences. We categorize noise into pointwise noise, which
includes low-quality data points, and pairwise noise, which encompasses
erroneous data pair associations that affect preference rankings. Utilizing
Distributionally Robust Optimization (DRO), we enhance DPO's resilience to
these types of noise. Our theoretical insights reveal that DPO inherently
embeds DRO principles, conferring robustness to pointwise noise, with the
regularization coefficient $\beta$ playing a critical role in its noise
resistance. Extending this framework, we introduce Distributionally
Robustifying DPO (Dr. DPO), which integrates pairwise robustness by optimizing
against worst-case pairwise scenarios. The novel hyperparameter $\beta'$ in Dr.
DPO allows for fine-tuned control over data pair reliability, providing a
strategic balance between exploration and exploitation in noisy training
environments. Empirical evaluations demonstrate that Dr. DPO substantially
improves the quality of generated text and response accuracy in preference
datasets, showcasing enhanced performance in both noisy and noise-free
settings. The code is available at https://github.com/junkangwu/Dr_DPO.
| [
{
"version": "v1",
"created": "Wed, 10 Jul 2024 17:48:25 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 08:05:53 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Wu",
"Junkang",
""
],
[
"Xie",
"Yuexiang",
""
],
[
"Yang",
"Zhengyi",
""
],
[
"Wu",
"Jiancan",
""
],
[
"Chen",
"Jiawei",
""
],
[
"Gao",
"Jinyang",
""
],
[
"Ding",
"Bolin",
""
],
[
"Wang",
"Xiang",
""
],
[
"He",
"Xiangnan",
""
]
] |
2407.07911 | Li Chen | Li Chen | A polynomial identity and beyond | null | null | null | null | math.NT math.FA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a novel polynomial identity and use it to prove an unexpected
result on linear algebra of quadratic forms. Besides its counterintuitive
nature which is of independent interest, our result happens to be closely
related to Druzkowski's reduction of the Jacobian Conjecture to which the
mystery can be partially attributed.
| [
{
"version": "v1",
"created": "Tue, 2 Jul 2024 12:35:16 GMT"
},
{
"version": "v2",
"created": "Thu, 8 Aug 2024 01:57:11 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 02:45:52 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Chen",
"Li",
""
]
] |
2407.08676 | Ahmad Omar . | Jiechao Feng and Ahmad K. Omar | Theory for the Anomalous Phase Behavior of Inertial Active Brownian
Particles | null | Phys. Rev. E 111, L043402 (2025) | 10.1103/PhysRevE.111.L043402 | null | cond-mat.soft cond-mat.stat-mech | http://creativecommons.org/licenses/by/4.0/ | In contrast to equilibrium systems, inertia can profoundly impact the phase
behavior of active systems. This has been made particularly evident in recent
years, with motility-induced phase separation (MIPS) exhibiting several
intriguing dependencies on translational inertia. Here we report extensive
simulations characterizing the phase behavior of inertial active matter and
develop a mechanical theory for the complete phase diagram without appealing to
equilibrium notions. Our theory qualitatively captures all aspects of
liquid-gas coexistence, including the critical value of inertia above which
MIPS ceases. Notably, our findings highlight that particle softness, and not
inertia, is responsible for the MIPS reentrance effect at the center of a
proposed active refrigeration cycle.
| [
{
"version": "v1",
"created": "Thu, 11 Jul 2024 17:09:23 GMT"
},
{
"version": "v2",
"created": "Mon, 30 Sep 2024 21:27:29 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 04:33:14 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Feng",
"Jiechao",
""
],
[
"Omar",
"Ahmad K.",
""
]
] |
2407.11969 | Maksym Andriushchenko | Maksym Andriushchenko and Nicolas Flammarion | Does Refusal Training in LLMs Generalize to the Past Tense? | Accepted at ICLR 2025. Updates in v2 and v3: added GPT-4o, Claude 3.5
Sonnet, o1-mini, and o1-preview results. Code and jailbreak artifacts:
https://github.com/tml-epfl/llm-past-tense | null | null | null | cs.CL cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Refusal training is widely used to prevent LLMs from generating harmful,
undesirable, or illegal outputs. We reveal a curious generalization gap in the
current refusal training approaches: simply reformulating a harmful request in
the past tense (e.g., "How to make a Molotov cocktail?" to "How did people make
a Molotov cocktail?") is often sufficient to jailbreak many state-of-the-art
LLMs. We systematically evaluate this method on Llama-3 8B, Claude-3.5 Sonnet,
GPT-3.5 Turbo, Gemma-2 9B, Phi-3-Mini, GPT-4o mini, GPT-4o, o1-mini,
o1-preview, and R2D2 models using GPT-3.5 Turbo as a reformulation model. For
example, the success rate of this simple attack on GPT-4o increases from 1%
using direct requests to 88% using 20 past tense reformulation attempts on
harmful requests from JailbreakBench with GPT-4 as a jailbreak judge.
Interestingly, we also find that reformulations in the future tense are less
effective, suggesting that refusal guardrails tend to consider past historical
questions more benign than hypothetical future questions. Moreover, our
experiments on fine-tuning GPT-3.5 Turbo show that defending against past
reformulations is feasible when past tense examples are explicitly included in
the fine-tuning data. Overall, our findings highlight that the widely used
alignment techniques -- such as SFT, RLHF, and adversarial training -- employed
to align the studied models can be brittle and do not always generalize as
intended. We provide code and jailbreak artifacts at
https://github.com/tml-epfl/llm-past-tense.
| [
{
"version": "v1",
"created": "Tue, 16 Jul 2024 17:59:55 GMT"
},
{
"version": "v2",
"created": "Fri, 19 Jul 2024 13:03:01 GMT"
},
{
"version": "v3",
"created": "Thu, 3 Oct 2024 16:46:09 GMT"
},
{
"version": "v4",
"created": "Thu, 17 Apr 2025 18:36:08 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Andriushchenko",
"Maksym",
""
],
[
"Flammarion",
"Nicolas",
""
]
] |
2407.14504 | Bahram Jalali | Callen MacPhee, Yiming Zhou, Bahram Jalali | Physical Data Embedding for Memory Efficient AI | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Deep neural networks (DNNs) have achieved exceptional performance across
various fields by learning complex, nonlinear mappings from large-scale
datasets. However, they face challenges such as high memory requirements and
computational costs with limited interpretability. This paper introduces an
approach where master equations of physics are converted into multilayered
networks that are trained via backpropagation. The resulting general-purpose
model effectively encodes data in the properties of the underlying physical
system. In contrast to existing methods wherein a trained neural network is
used as a computationally efficient alternative for solving physical equations,
our approach directly treats physics equations as trainable models. We
demonstrate this physical embedding concept with the Nonlinear Schr\"odinger
Equation (NLSE), which acts as trainable architecture for learning complex
patterns including nonlinear mappings and memory effects from data. The network
embeds data representation in orders of magnitude fewer parameters than
conventional neural networks when tested on time series data. Notably, the
trained "Nonlinear Schr\"odinger Network" is interpretable, with all parameters
having physical meanings. This interpretability offers insight into the
underlying dynamics of the system that produced the data. The proposed method
of replacing traditional DNN feature learning architectures with physical
equations is also extended to the Gross-Pitaevskii Equation, demonstrating the
broad applicability of the framework to other master equations of physics.
Among our results, an ablation study quantifies the relative importance of
physical terms such as dispersion, nonlinearity, and potential energy for
classification accuracy. We also outline the limitations of this approach as it
relates to generalizability.
| [
{
"version": "v1",
"created": "Fri, 19 Jul 2024 17:58:00 GMT"
},
{
"version": "v2",
"created": "Wed, 24 Jul 2024 04:33:55 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 00:29:41 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"MacPhee",
"Callen",
""
],
[
"Zhou",
"Yiming",
""
],
[
"Jalali",
"Bahram",
""
]
] |
2407.15289 | Marco Galoppo | Marco Galoppo, Rudeep Gaur and Christopher Harvey-Hawes | Kerr--Newman Memory Effect | 24 pages, 0 figures | Phys. Rev. D 111, 064041 (2025) | 10.1103/PhysRevD.111.064041 | null | gr-qc hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We bring the Kerr--Newman spacetime into the Bondi--Sachs gauge by means of
zero angular momentum, null geodesics. We compute the memory effect produced at
the black hole horizon by a transient gravitational shock wave, which from
future null infinity is seen as a Bondi-Metzner-Sachs supertranslation. This
results in a change of the supertransformation charges at infinity between the
spacetime geometries defined by the black hole before, and after, the shockwave
scattering. For an extremal Kerr--Newman black hole, we give the complementary
description of this process in the near-horizon limit, as seen by an observer
hovering over the horizon. In this limit, we compute the supertranformation
charges and compare them to those calculated at null infinity. We analyze the
effect of these transformations on the electromagnetic gauge field and explore
the self-interaction between this and the angular momentum of the black hole.
| [
{
"version": "v1",
"created": "Sun, 21 Jul 2024 22:56:38 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Galoppo",
"Marco",
""
],
[
"Gaur",
"Rudeep",
""
],
[
"Harvey-Hawes",
"Christopher",
""
]
] |
2407.17763 | Xiaofeng Xu | Jinchao Xu, Xiaofeng Xu | Randomized Greedy Algorithms for Neural Network Optimization | 25 pages, 7 figures | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Greedy algorithms have been successfully analyzed and applied in training
neural networks for solving variational problems, ensuring guaranteed
convergence orders. In this paper, we extend the analysis of the orthogonal
greedy algorithm (OGA) to convex optimization problems, establishing its
optimal convergence rate. This result broadens the applicability of OGA by
generalizing its optimal convergence rate from function approximation to convex
optimization problems. In addition, we also address the issue regarding
practical applicability of greedy algorithms, which is due to significant
computational costs from the subproblems that involve an exhaustive search over
a discrete dictionary. We propose to use a more practical approach of randomly
discretizing the dictionary at each iteration of the greedy algorithm. We
quantify the required size of the randomized discrete dictionary and prove
that, with high probability, the proposed algorithm realizes a weak greedy
algorithm, achieving optimal convergence orders. Through numerous numerical
experiments on function approximation, linear and nonlinear elliptic partial
differential equations, we validate our analysis on the optimal convergence
rate and demonstrate the advantage of using randomized discrete dictionaries
over a deterministic one by showing orders of magnitude reductions in the size
of the discrete dictionary, particularly in higher dimensions.
| [
{
"version": "v1",
"created": "Thu, 25 Jul 2024 04:38:17 GMT"
},
{
"version": "v2",
"created": "Sun, 15 Sep 2024 08:22:10 GMT"
},
{
"version": "v3",
"created": "Thu, 17 Apr 2025 21:50:28 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Xu",
"Jinchao",
""
],
[
"Xu",
"Xiaofeng",
""
]
] |
2407.17908 | Ivo Souza | Ivo Souza, Richard M. Martin, Massimiliano Stengel | Optical bounds on many-electron localization | null | SciPost Phys. 18, 127 (2025) | 10.21468/SciPostPhys.18.4.127 | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We establish rigorous inequalities between different electronic properties
linked to optical sum rules, and organize them into weak and strong bounds on
three characteristic properties of insulators: electron localization length
$\ell$ (the quantum fluctuations in polarization), electric susceptibility
$\chi$, and optical gap $E_{\rm G}$. All-electron and valence-only versions of
the bounds are given, and the latter are found to be more informative. The
bounds on $\ell$ are particularly interesting, as they provide reasonably tight
estimates for an ellusive ground-state property - the average localization
length of valence electrons - from tabulated experimental data: electron
density, high-frequency dielectric constant, and optical gap. The localization
lengths estimated in this way for several materials follow simple chemical
trends, especially for the alkali halides. We also illustrate our findings via
analytically solvable harmonic oscillator models, which reveal an intriguing
connection to the physics of long-ranged van der Waals forces.
| [
{
"version": "v1",
"created": "Thu, 25 Jul 2024 09:53:30 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 10:38:46 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Souza",
"Ivo",
""
],
[
"Martin",
"Richard M.",
""
],
[
"Stengel",
"Massimiliano",
""
]
] |
2407.18345 | Taras Radul | Taras Radul | On the functor of comonotonically maxitive functionals | arXiv admin note: text overlap with arXiv:2307.12787,
arXiv:2111.06612 | Topology and its Applications , 2024, 358, 109131 | null | null | math.GN | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We introduce a functor of functionals which preserve maximum of comonotone
functions and addition of constants. This functor is a subfunctor of the
functor of order-preserving functionals and contains the idempotent measure
functor as subfunctor. The main aim of this paper is to show that this functor
is isomorphic to the capacity functor. We establish such isomorphism using the
fuzzy max-plus integral. In fact, we can consider this result as an idempotent
analogue of Riesz Theorem about a correspondence between the set of
$\sigma$-additive regular Borel measures and the set of linear positively
defined functionals.
| [
{
"version": "v1",
"created": "Thu, 25 Jul 2024 19:18:20 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Radul",
"Taras",
""
]
] |
2407.19611 | Tsung-Sheng Huang | Tsung-Sheng Huang, Yu-Xin Wang, Yan-Qi Wang, Darrick Chang, Mohammad
Hafezi, Andrey Grankin | Collective optical properties of moir\'e excitons | null | null | null | null | cond-mat.mes-hall cond-mat.mtrl-sci cond-mat.str-el quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose that excitons in moir\'e transition metal dichalcogenide bilayers
offer a promising platform for investigating collective radiative properties.
While some of these optical properties resemble those of cold atom arrays,
moir\'e excitons extend to the deep subwavelength limit, beyond the reach of
current optical lattice experiments. Remarkably, we show that the collective
optical properties can be exploited to probe certain correlated electron states
without requiring subwavelength spatial resolution. Specifically, we illustrate
that the Wigner crystal states of electrons doped into these bilayers act as an
emergent periodic potential for excitons. Moreover, the collective dissipative
excitonic bands and their associated Berry curvature can reveal various charge
orders that emerge at the corresponding electronic doping. Our study provides a
promising pathway for future research on the interplay between collective
effects and strong correlations involving moir\'e excitons.
| [
{
"version": "v1",
"created": "Sun, 28 Jul 2024 23:43:46 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 15:38:23 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Huang",
"Tsung-Sheng",
""
],
[
"Wang",
"Yu-Xin",
""
],
[
"Wang",
"Yan-Qi",
""
],
[
"Chang",
"Darrick",
""
],
[
"Hafezi",
"Mohammad",
""
],
[
"Grankin",
"Andrey",
""
]
] |
2407.20500 | Tarun Grover | Ting-Tung Wang, Menghan Song, Zi Yang Meng, Tarun Grover | An analog of topological entanglement entropy for mixed states | (Accepted version) 16 pages main text, 3 pages of appendices, 7
figures, minor generalizations added to Sec.III, typos fixed | PRX Quantum 6, 010358 (2025) | 10.1103/PRXQuantum.6.010358 | null | quant-ph cond-mat.stat-mech cond-mat.str-el hep-th | http://creativecommons.org/licenses/by/4.0/ | We propose the convex-roof extension of quantum conditional mutual
information ("co(QCMI)") as a diagnostic of topological order in a mixed state.
We focus primarily on topological states subjected to local decoherence, and
employ the Levin-Wen scheme to define co(QCMI), so that for a pure state,
co(QCMI) equals topological entanglement entropy (TEE). By construction,
co(QCMI) is zero if and only if a mixed state can be decomposed as a convex sum
of pure states with zero TEE. We show that co(QCMI) is non-increasing with
increasing decoherence when Kraus operators are proportional to the product of
onsite unitaries. This implies that unlike a pure state transition between a
topologically trivial and a non-trivial phase, the long-range entanglement at a
decoherence-induced topological phase transition as quantified by co(QCMI) is
less than or equal to that in the proximate topological phase. For the 2d toric
code decohered by onsite bit/phase-flip noise, we show that co(QCMI) is
non-zero below the error-recovery threshold and zero above it. Relatedly, the
decohered state cannot be written as a convex sum of short-range entangled pure
states below the threshold. We conjecture and provide evidence that in this
example, co(QCMI) equals TEE of a recently introduced pure state. In
particular, we develop a tensor-assisted Monte Carlo (TMC) computation method
to efficiently evaluate the R\'enyi TEE for the aforementioned pure state and
provide non-trivial consistency checks for our conjecture. We use TMC to also
calculate the universal scaling dimension of the anyon-condensation order
parameter at this transition.
| [
{
"version": "v1",
"created": "Tue, 30 Jul 2024 02:26:45 GMT"
},
{
"version": "v2",
"created": "Fri, 21 Mar 2025 00:08:04 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Wang",
"Ting-Tung",
""
],
[
"Song",
"Menghan",
""
],
[
"Meng",
"Zi Yang",
""
],
[
"Grover",
"Tarun",
""
]
] |
2407.20751 | Hidekazu Yoshioka | Hidekazu Yoshioka | Generalized replicator dynamics based on mean-field pairwise comparison
dynamic | A preprint of some submitted manuscript | null | 10.1016/j.matcom.2025.04.010 | null | math.OC math.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The pairwise comparison dynamic is a forward ordinary differential equation
in a Banach space whose solution is a time-dependent probability measure to
maximize utility based on a nonlinear and nonlocal protocol. It contains a wide
class of evolutionary game models, such as replicator dynamics and its
generalization. We present an inverse control approach to obtain a
replicator-type pairwise comparison dynamic from the large discount limit of a
mean field game (MFG) as a coupled forward-backward system. This methodology
provides a new interpretation of replicator-type dynamics as a myopic
perception limit of the dynamic programming. The cost function in the MFG is
explicitly obtained to derive the generalized replicator dynamics. We present a
finite difference method to compute these models such that the conservation and
nonnegativity of the probability density and bounds of the value function can
be numerically satisfied. We conduct a computational convergence study of a
large discount limit, focusing on potential games and an energy management
problem under several conditions.
| [
{
"version": "v1",
"created": "Tue, 30 Jul 2024 11:48:42 GMT"
},
{
"version": "v2",
"created": "Sun, 2 Feb 2025 04:14:08 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Yoshioka",
"Hidekazu",
""
]
] |
2407.20999 | Senmiao Wang | Yupeng Chen, Senmiao Wang, Yushun Zhang, Zhihang Lin, Haozhe Zhang,
Weijian Sun, Tian Ding, Ruoyu Sun | MoFO: Momentum-Filtered Optimizer for Mitigating Forgetting in LLM
Fine-Tuning | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) have demonstrated remarkable capabilities across
a wide range of tasks. Typically, LLMs are first pre-trained on large corpora
and subsequently fine-tuned on task-specific datasets. However, during
fine-tuning, LLMs may forget some knowledge acquired in the pre-training stage,
leading to a decline in general capabilities. Existing approaches to mitigate
forgetting often rely on access to pre-training data, which may be unavailable
in many real-world scenarios--such as fine-tuning checkpoint-only open-source
LLMs. To address this challenge, we propose a new fine-tuning algorithm termed
Momentum-Filtered Optimizer (MoFO). MoFO is an extension of greedy block
coordinate descent (BCD) methods: in each iteration, MoFO only updates the
model parameters with the largest momentum magnitudes, while keeping all other
parameters fixed. MoFO achieves similar fine-tuning performance to the default
fine-tuning algorithm while effectively mitigating knowledge forgetting. We
validate MoFO through rigorous convergence analysis and extensive experiments,
demonstrating its effectiveness in mitigating forgetting without pre-training
data.
| [
{
"version": "v1",
"created": "Tue, 30 Jul 2024 17:38:24 GMT"
},
{
"version": "v2",
"created": "Wed, 31 Jul 2024 17:56:03 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 09:04:15 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Chen",
"Yupeng",
""
],
[
"Wang",
"Senmiao",
""
],
[
"Zhang",
"Yushun",
""
],
[
"Lin",
"Zhihang",
""
],
[
"Zhang",
"Haozhe",
""
],
[
"Sun",
"Weijian",
""
],
[
"Ding",
"Tian",
""
],
[
"Sun",
"Ruoyu",
""
]
] |
2408.02288 | Haiping Huang | Yuhao Li, Ruoran Bai, Haiping Huang | Spin glass model of in-context learning | 16 pages, 4+6 figures, revised version to the journal | null | null | null | cond-mat.dis-nn cond-mat.stat-mech cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | Large language models show a surprising in-context learning ability -- being
able to use a prompt to form a prediction for a query, yet without additional
training, in stark contrast to old-fashioned supervised learning. Providing a
mechanistic interpretation and linking the empirical phenomenon to physics are
thus challenging and remain unsolved. We study a simple yet expressive
transformer with linear attention and map this structure to a spin glass model
with real-valued spins, where the couplings and fields explain the intrinsic
disorder in data. The spin glass model explains how the weight parameters
interact with each other during pre-training, and further clarifies why an
unseen function can be predicted by providing only a prompt yet without further
training. Our theory reveals that for single-instance learning, increasing the
task diversity leads to the emergence of in-context learning, by allowing the
Boltzmann distribution to converge to a unique correct solution of weight
parameters. Therefore the pre-trained transformer displays a prediction power
in a novel prompt setting. The proposed analytically tractable model thus
offers a promising avenue for thinking about how to interpret many intriguing
but puzzling properties of large language models.
| [
{
"version": "v1",
"created": "Mon, 5 Aug 2024 07:54:01 GMT"
},
{
"version": "v2",
"created": "Wed, 13 Nov 2024 07:13:36 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 08:16:22 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Li",
"Yuhao",
""
],
[
"Bai",
"Ruoran",
""
],
[
"Huang",
"Haiping",
""
]
] |
2408.02657 | Le Zhuo | Dongyang Liu, Shitian Zhao, Le Zhuo, Weifeng Lin, Yu Qiao, Hongsheng
Li, Peng Gao | Lumina-mGPT: Illuminate Flexible Photorealistic Text-to-Image Generation
with Multimodal Generative Pretraining | Code available at: https://github.com/Alpha-VLLM/Lumina-mGPT | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | We present Lumina-mGPT, a family of multimodal autoregressive models capable
of various vision and language tasks, particularly excelling in generating
flexible photorealistic images from text descriptions. By initializing from
multimodal Generative PreTraining (mGPT), we demonstrate that decoder-only
Autoregressive (AR) model can achieve image generation performance comparable
to modern diffusion models with high efficiency through Flexible Progressive
Supervised Fine-tuning (FP-SFT). Equipped with our proposed Unambiguous image
Representation (UniRep), Lumina-mGPT can flexibly generate high-quality images
of varying aspect ratios. Building on the strong image generation capabilities,
we further explore Ominiponent Supervised Fine-tuning (Omni-SFT), an initial
attempt to elevate Lumina-mGPT into a unified multi-modal generalist. The
resulting model demonstrates versatile multimodal capabilities, including
visual generation tasks like text-to-image/multiview generation and
controllable generation, visual recognition tasks like segmentation and depth
estimation, and vision-language tasks like multi-turn visual question
answering, showing the rosy potential of the technical direction. Codes and
checkpoints are available at https://github.com/Alpha-VLLM/Lumina-mGPT.
| [
{
"version": "v1",
"created": "Mon, 5 Aug 2024 17:46:53 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 15:32:16 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Liu",
"Dongyang",
""
],
[
"Zhao",
"Shitian",
""
],
[
"Zhuo",
"Le",
""
],
[
"Lin",
"Weifeng",
""
],
[
"Qiao",
"Yu",
""
],
[
"Li",
"Hongsheng",
""
],
[
"Gao",
"Peng",
""
]
] |
2408.03131 | Ming Chenlin | Chenlin Ming, Zitong Wang, Boxuan Zhang, Zhanxiang Cao, Xiaoming Duan,
Jianping He | Stochastic Trajectory Optimization for Robotic Skill Acquisition From a
Suboptimal Demonstration | null | null | null | null | cs.RO cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Learning from Demonstration (LfD) has emerged as a crucial method for robots
to acquire new skills. However, when given suboptimal task trajectory
demonstrations with shape characteristics reflecting human preferences but
subpar dynamic attributes such as slow motion, robots not only need to mimic
the behaviors but also optimize the dynamic performance. In this work, we
leverage optimization-based methods to search for a superior-performing
trajectory whose shape is similar to that of the demonstrated trajectory.
Specifically, we use Dynamic Time Warping (DTW) to quantify the difference
between two trajectories and combine it with additional performance metrics,
such as collision cost, to construct the cost function. Moreover, we develop a
multi-policy version of the Stochastic Trajectory Optimization for Motion
Planning (STOMP), called MSTOMP, which is more stable and robust to parameter
changes. To deal with the jitter in the demonstrated trajectory, we further
utilize the gain-controlling method in the frequency domain to denoise the
demonstration and propose a computationally more efficient metric, called Mean
Square Error in the Spectrum (MSES), that measures the trajectories'
differences in the frequency domain. We also theoretically highlight the
connections between the time domain and the frequency domain methods. Finally,
we verify our method in both simulation experiments and real-world experiments,
showcasing its improved optimization performance and stability compared to
existing methods.
| [
{
"version": "v1",
"created": "Tue, 6 Aug 2024 12:16:15 GMT"
},
{
"version": "v2",
"created": "Wed, 7 Aug 2024 02:34:32 GMT"
},
{
"version": "v3",
"created": "Wed, 16 Apr 2025 14:15:28 GMT"
},
{
"version": "v4",
"created": "Fri, 18 Apr 2025 05:47:23 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Ming",
"Chenlin",
""
],
[
"Wang",
"Zitong",
""
],
[
"Zhang",
"Boxuan",
""
],
[
"Cao",
"Zhanxiang",
""
],
[
"Duan",
"Xiaoming",
""
],
[
"He",
"Jianping",
""
]
] |
2408.04342 | Paul R. B. Houssel | Paul R. B. Houssel and Priyanka Singh and Siamak Layeghy and Marius
Portmann | Towards Explainable Network Intrusion Detection using Large Language
Models | null | null | 10.1109/BDCAT63179.2024.00021 | null | cs.CR cs.AI cs.NI | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) have revolutionised natural language processing
tasks, particularly as chat agents. However, their applicability to threat
detection problems remains unclear. This paper examines the feasibility of
employing LLMs as a Network Intrusion Detection System (NIDS), despite their
high computational requirements, primarily for the sake of explainability.
Furthermore, considerable resources have been invested in developing LLMs, and
they may offer utility for NIDS. Current state-of-the-art NIDS rely on
artificial benchmarking datasets, resulting in skewed performance when applied
to real-world networking environments. Therefore, we compare the GPT-4 and
LLama3 models against traditional architectures and transformer-based models to
assess their ability to detect malicious NetFlows without depending on
artificially skewed datasets, but solely on their vast pre-trained acquired
knowledge. Our results reveal that, although LLMs struggle with precise attack
detection, they hold significant potential for a path towards explainable NIDS.
Our preliminary exploration shows that LLMs are unfit for the detection of
Malicious NetFlows. Most promisingly, however, these exhibit significant
potential as complementary agents in NIDS, particularly in providing
explanations and aiding in threat response when integrated with Retrieval
Augmented Generation (RAG) and function calling capabilities.
| [
{
"version": "v1",
"created": "Thu, 8 Aug 2024 09:59:30 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Houssel",
"Paul R. B.",
""
],
[
"Singh",
"Priyanka",
""
],
[
"Layeghy",
"Siamak",
""
],
[
"Portmann",
"Marius",
""
]
] |
2408.04820 | Kensen Shi | Kensen Shi, Deniz Alt{\i}nb\"uken, Saswat Anand, Mihai Christodorescu,
Katja Gr\"unwedel, Alexa Koenings, Sai Naidu, Anurag Pathak, Marc Rasi,
Fredde Ribeiro, Brandon Ruffin, Siddhant Sanyam, Maxim Tabachnyk, Sara Toth,
Roy Tu, Tobias Welp, Pengcheng Yin, Manzil Zaheer, Satish Chandra, Charles
Sutton | Natural Language Outlines for Code: Literate Programming in the LLM Era | Accepted to FSE'25 Industry Track | null | 10.1145/3696630.3728541 | null | cs.SE cs.AI cs.CL cs.HC cs.LG | http://creativecommons.org/licenses/by/4.0/ | We propose using natural language outlines as a novel modality and
interaction surface for providing AI assistance to developers throughout the
software development process. An NL outline for a code function comprises
multiple statements written in concise prose, which partition the code and
summarize its main ideas in the style of literate programming. Crucially, we
find that modern LLMs can generate accurate and high-quality NL outlines in
practice. Moreover, NL outlines enable a bidirectional sync between code and
NL, where a developer can change either code or NL and have the LLM
automatically update the other. We discuss many use cases for NL outlines: they
can accelerate understanding and navigation of code and diffs, simplify code
maintenance, augment code search, steer code generation, and more. We then
propose and compare multiple LLM prompting techniques for generating outlines
and ask professional developers to judge outline quality. Finally, we present
two case studies applying NL outlines toward code review and malware detection.
| [
{
"version": "v1",
"created": "Fri, 9 Aug 2024 02:22:51 GMT"
},
{
"version": "v2",
"created": "Wed, 15 Jan 2025 03:43:22 GMT"
},
{
"version": "v3",
"created": "Wed, 16 Apr 2025 07:09:59 GMT"
},
{
"version": "v4",
"created": "Thu, 17 Apr 2025 22:02:36 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Shi",
"Kensen",
""
],
[
"Altınbüken",
"Deniz",
""
],
[
"Anand",
"Saswat",
""
],
[
"Christodorescu",
"Mihai",
""
],
[
"Grünwedel",
"Katja",
""
],
[
"Koenings",
"Alexa",
""
],
[
"Naidu",
"Sai",
""
],
[
"Pathak",
"Anurag",
""
],
[
"Rasi",
"Marc",
""
],
[
"Ribeiro",
"Fredde",
""
],
[
"Ruffin",
"Brandon",
""
],
[
"Sanyam",
"Siddhant",
""
],
[
"Tabachnyk",
"Maxim",
""
],
[
"Toth",
"Sara",
""
],
[
"Tu",
"Roy",
""
],
[
"Welp",
"Tobias",
""
],
[
"Yin",
"Pengcheng",
""
],
[
"Zaheer",
"Manzil",
""
],
[
"Chandra",
"Satish",
""
],
[
"Sutton",
"Charles",
""
]
] |
2408.05350 | Landon Dyken | Landon Dyken, Saugat Adhikari, Pravin Poudel, Steve Petruzza, Da Yan,
Will Usher, Sidharth Kumar | Enabling Fast and Accurate Crowdsourced Annotation for Elevation-Aware
Flood Extent Mapping | null | null | null | null | cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Mapping the extent of flood events is a necessary and important aspect of
disaster management. In recent years, deep learning methods have evolved as an
effective tool to quickly label high resolution imagery and provide necessary
flood extent mappings. These methods, though, require large amounts of
annotated training data to create models that are accurate and robust to new
flooded imagery. In this work, we present FloodTrace, a web-based application
that enables effective crowdsourcing of flooded region annotation for machine
learning applications. To create this application, we conducted extensive
interviews with domain experts to produce a set of formal requirements. Our
work brings topological segmentation tools to the web and greatly improves
annotation efficiency compared to the state-of-the-art. The user-friendliness
of our solution allows researchers to outsource annotations to non-experts and
utilize them to produce training data with equal quality to fully
expert-labeled data. We conducted a user study to confirm the effectiveness of
our application in which 266 graduate students annotated high-resolution aerial
imagery from Hurricane Matthew in North Carolina. Experimental results show the
efficiency benefits of our application for untrained users, with median
annotation time less than half the state-of-the-art annotation method. In
addition, using our aggregation and correction framework, flood detection
models trained on crowdsourced annotations were able to achieve performance
equal to models trained on fully expert-labeled annotations, while requiring a
fraction of the time on the part of the expert.
| [
{
"version": "v1",
"created": "Wed, 31 Jul 2024 23:42:05 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Apr 2025 18:53:23 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Dyken",
"Landon",
""
],
[
"Adhikari",
"Saugat",
""
],
[
"Poudel",
"Pravin",
""
],
[
"Petruzza",
"Steve",
""
],
[
"Yan",
"Da",
""
],
[
"Usher",
"Will",
""
],
[
"Kumar",
"Sidharth",
""
]
] |
2408.06896 | Yurii Ignat'ev | Yu. G. Ignat'ev | Evolution of plane perturbations in the cosmological environment of the
Higgs scalar field and an ideal scalar charged fluid | 26 pages, 24 figures, 26 references | Theor. Math. Phys., 222, 285 (2025) | 10.1134/S0040577925020072 | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A phenomenological model of an ideal fluid with a scalar charge is
formulated, on the basis of which a model with a neutral fluid and a
vacuum-field model with rules of transition between them are constructed. A
qualitative analysis of the obtained dynamic systems is carried out and
numerical cosmological models based on these systems are constructed. A
mathematical model of plane longitudinal scalar-gravitational perturbations of
the Friedmann ideal charged fluid with Higgs interaction is formulated. It is
shown that in the absence of fluid, i.e., in the vacuum-field model,
gravitational perturbations do not arise. Perturbations of the scalar field are
possible only in those cases when in the unperturbed state the cosmological
system is at singular points. For these cases, exact solutions of the field
equation are found, expressed in Bessel functions of the first and second kind
and describing damped oscillations in the case of a stable unperturbed state
and growing oscillations in the case of an unstable unperturbed state. The WKB
theory of plane scalar-gravitational perturbations is constructed: dispersion
equations are obtained in general form and solved for a neutral fluid. In this
case, expressions are obtained for the local frequency and growth increment of
oscillations, as well as the integral increment. It is shown that only free
wave regimes or growing standing oscillations are possible during the
evolution. Perturbations in the WKB approximation in a neutral fluid are
studied and it is shown that local formulas for the evolution of perturbations
correspond to the model of the 1985 article by M.Yu. The times of the beginning
and end of the instability phase are determined and it is shown that
instability can develop only at the unstable inflationary stage of the
expansion of the Universe.
| [
{
"version": "v1",
"created": "Tue, 13 Aug 2024 13:44:09 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Ignat'ev",
"Yu. G.",
""
]
] |
2408.08070 | Fenghe Tang | Fenghe Tang, Bingkun Nian, Yingtai Li, Zihang Jiang, Jie Yang, Wei
Liu, S. Kevin Zhou | MambaMIM: Pre-training Mamba with State Space Token Interpolation and
its Application to Medical Image Segmentation | Accepted by Medical Image Analysis. Code:
https://github.com/FengheTan9/MambaMIM | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, the state space model Mamba has demonstrated efficient
long-sequence modeling capabilities, particularly for addressing long-sequence
visual tasks in 3D medical imaging. However, existing generative
self-supervised learning methods have not yet fully unleashed Mamba's potential
for handling long-range dependencies because they overlook the inherent causal
properties of state space sequences in masked modeling. To address this
challenge, we propose a general-purpose pre-training framework called MambaMIM,
a masked image modeling method based on a novel TOKen-Interpolation strategy
(TOKI) for the selective structure state space sequence, which learns causal
relationships of state space within the masked sequence. Further, MambaMIM
introduces a bottom-up 3D hybrid masking strategy to maintain a masking
consistency across different architectures and can be used on any single or
hybrid Mamba architecture to enhance its multi-scale and long-range
representation capability. We pre-train MambaMIM on a large-scale dataset of
6.8K CT scans and evaluate its performance across eight public medical
segmentation benchmarks. Extensive downstream experiments reveal the
feasibility and advancement of using Mamba for medical image pre-training. In
particular, when we apply the MambaMIM to a customized architecture that
hybridizes MedNeXt and Vision Mamba, we consistently obtain the
state-of-the-art segmentation performance. The code is available at:
https://github.com/FengheTan9/MambaMIM.
| [
{
"version": "v1",
"created": "Thu, 15 Aug 2024 10:35:26 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 13:21:39 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Tang",
"Fenghe",
""
],
[
"Nian",
"Bingkun",
""
],
[
"Li",
"Yingtai",
""
],
[
"Jiang",
"Zihang",
""
],
[
"Yang",
"Jie",
""
],
[
"Liu",
"Wei",
""
],
[
"Zhou",
"S. Kevin",
""
]
] |
2408.08304 | Friederike Becker | Friederike Becker, Fabian Kr\"uger, Melanie Schienle | Simple Macroeconomic Forecast Distributions for the G7 Economies | null | null | null | null | stat.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a simple method for predicting the distribution of output growth
and inflation in the G7 economies. The method is based on point forecasts
published by the International Monetary Fund (IMF), as well as robust
statistics from the empirical distribution of the IMF's past forecast errors
while imposing coherence of prediction intervals across horizons. We show that
the technique yields calibrated prediction intervals and performs similar to,
or better than, more complex time series models in terms of statistical loss
functions. We provide a simple website with graphical illustrations of our
forecasts, as well as time-stamped data files that document their real time
character.
| [
{
"version": "v1",
"created": "Thu, 15 Aug 2024 17:57:25 GMT"
},
{
"version": "v2",
"created": "Wed, 25 Sep 2024 17:59:30 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 17:44:58 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Becker",
"Friederike",
""
],
[
"Krüger",
"Fabian",
""
],
[
"Schienle",
"Melanie",
""
]
] |
2408.09967 | Andrew Kiruluta | Andrew Kiruluta, Andreas Lemos | Unsupervised Machine Learning Hybrid Approach Integrating Linear
Programming in Loss Function: A Robust Optimization Technique | null | null | null | null | cs.LG cs.AI math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a novel hybrid approach that integrates linear
programming (LP) within the loss function of an unsupervised machine learning
model. By leveraging the strengths of both optimization techniques and machine
learning, this method introduces a robust framework for solving complex
optimization problems where traditional methods may fall short. The proposed
approach encapsulates the constraints and objectives of a linear programming
problem directly into the loss function, guiding the learning process to adhere
to these constraints while optimizing the desired outcomes. This technique not
only preserves the interpretability of linear programming but also benefits
from the flexibility and adaptability of machine learning, making it
particularly well-suited for unsupervised or semi-supervised learning
scenarios.
| [
{
"version": "v1",
"created": "Mon, 19 Aug 2024 13:14:26 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 14:16:20 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Kiruluta",
"Andrew",
""
],
[
"Lemos",
"Andreas",
""
]
] |
2408.10979 | Hongfa Zi | Hongfa Zi, Zhen Liu | The fusion of phonography and ideographic characters into virtual
Chinese characters -- Based on Chinese and English | 14 pages, 7 figures | null | null | null | cs.OH | http://creativecommons.org/licenses/by/4.0/ | The characters used in modern countries are mainly divided into ideographic
characters and phonetic characters, both of which have their advantages and
disadvantages. Chinese is difficult to learn and easy to master, while English
is easy to learn but has a large vocabulary. There is still no language that
combines the advantages of both languages and has less memory capacity, can
form words, and is easy to learn. Therefore, inventing new characters that can
be combined and the popularization of deep knowledge, and reduce disputes
through communication. Firstly, observe the advantages and disadvantages of
Chinese and English, such as their vocabulary, information content, and ease of
learning in deep scientific knowledge, and create a new writing system. Then,
use comparative analysis to observe the total score of the new language.
Through this article, it can be concluded that the new text combines the
advantages of both pictographic and alphabetical writing: new characters that
can be combined into words reduces the vocabulary that needs to be learned;
Special prefixes allow beginners to quickly guess the approximate category and
meaning of unseen words; New characters can enable humans to quickly learn more
advanced knowledge.
| [
{
"version": "v1",
"created": "Tue, 20 Aug 2024 16:13:28 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Zi",
"Hongfa",
""
],
[
"Liu",
"Zhen",
""
]
] |
2408.11317 | Hongfa Zi | Hongfa Zi, Zhen Liu | The continuous accumulation of civilization core in the cycle of
elements-creature, benefits and weapons | 14 pages, 7 figures | null | null | null | cs.OH | http://creativecommons.org/licenses/by/4.0/ | The comprehensive strength of a country varies from strong to weak, divided
into three condition: descending, periodicity destruction or rapidly rising,
Exploring the differences can solve the development crisis. the most important
things for a country are interests, weapons and creature, corresponding to
money, technology and people. The ship industry has two attribute of financial
benefits and technological weapons. Commercial ships can transport massive
commodity and warships carry updating of massive technological weapons; But a
new core: equity incentives have emerged, and it has helped the rapid
development of the computer industry. This article uses comparative analysis
and comparative historical analysis to observe the changes in the United States
and China after the mutual circulation of two elements and the double
circulation of three elements in history, such as the growth rates of GDP and
patent applications. Then, it summarizes the changes brought by the core of
civilization to the country.Through this article, it can be concluded that the
core of civilization consists of ships, equity incentives, record-wisdom
method; Through the circulation of new elements, a country can transform into
civilizations with three cycles, achieving mutual circulation among the three
and enhancing endogenous power; The core of civilization can enhance the
stability of economic development, prevent economic crises, and achieve a more
balanced civilization.
| [
{
"version": "v1",
"created": "Wed, 21 Aug 2024 03:51:40 GMT"
},
{
"version": "v2",
"created": "Fri, 30 Aug 2024 00:31:05 GMT"
},
{
"version": "v3",
"created": "Wed, 9 Oct 2024 14:29:22 GMT"
},
{
"version": "v4",
"created": "Thu, 17 Oct 2024 19:32:24 GMT"
},
{
"version": "v5",
"created": "Tue, 5 Nov 2024 13:27:27 GMT"
},
{
"version": "v6",
"created": "Thu, 20 Mar 2025 07:44:03 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Zi",
"Hongfa",
""
],
[
"Liu",
"Zhen",
""
]
] |
2408.11363 | Yijia Xiao | Yijia Xiao, Edward Sun, Yiqiao Jin, Qifan Wang, Wei Wang | ProteinGPT: Multimodal LLM for Protein Property Prediction and Structure
Understanding | Spotlight, Machine Learning for Genomics Explorations @ ICLR 2025 | null | null | null | cs.AI cs.CE cs.LG q-bio.BM | http://creativecommons.org/licenses/by/4.0/ | Understanding biological processes, drug development, and biotechnological
advancements requires a detailed analysis of protein structures and functions,
a task that is inherently complex and time-consuming in traditional protein
research. To streamline this process, we introduce ProteinGPT, a
state-of-the-art multimodal large language model for proteins that enables
users to upload protein sequences and/or structures for comprehensive analysis
and responsive inquiries. ProteinGPT integrates protein sequence and structure
encoders with linear projection layers to ensure precise representation
adaptation and leverages a large language model (LLM) to generate accurate,
contextually relevant responses. To train ProteinGPT, we constructed a
large-scale dataset of 132,092 proteins, each annotated with 20-30 property
tags and 5-10 QA pairs per protein, and optimized the instruction-tuning
process using GPT-4o. Experiments demonstrate that ProteinGPT effectively
generates informative responses to protein-related questions, achieving high
performance on both semantic and lexical metrics and significantly
outperforming baseline models and general-purpose LLMs in understanding and
responding to protein-related queries. Our code and data are available at
https://github.com/ProteinGPT/ProteinGPT.
| [
{
"version": "v1",
"created": "Wed, 21 Aug 2024 06:16:22 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Apr 2025 21:55:37 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Xiao",
"Yijia",
""
],
[
"Sun",
"Edward",
""
],
[
"Jin",
"Yiqiao",
""
],
[
"Wang",
"Qifan",
""
],
[
"Wang",
"Wei",
""
]
] |
2408.11904 | Gonzalo Herrera | Gonzalo Herrera and Patrick Huber | Anapole moment of neutrinos and radioactive sources near liquid xenon
detectors | Matches version published in PRD | Phys. Rev. D 111, 073008, 2025 | 10.1103/PhysRevD.111.073008 | null | hep-ph hep-ex | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that placing a radioactive source such as ${}^{51}$Cr near a liquid
xenon detector may allow to detect the contribution induced by the anapole
moment to neutrino-electron scattering in the Standard Model at the 1-2
$\sigma$ level. Although the anapole moment of neutrinos induces a scattering
rate with the same spectral shape as the neutral and charged current
contributions, exposures of $\sim$ 60 tonne $\times$ source run at XENONnT or
XLZD may be enough to accumulate sufficient statistics for a detection. We also
discuss a simple model where the anapole moment of neutrinos is enhanced or
decreased with respect to the SM expectation, further demonstrating how a
potential measurement of the anapole moment of neutrinos would allow to
constrain new physics.
| [
{
"version": "v1",
"created": "Wed, 21 Aug 2024 18:00:23 GMT"
},
{
"version": "v2",
"created": "Mon, 30 Sep 2024 21:50:42 GMT"
},
{
"version": "v3",
"created": "Fri, 18 Apr 2025 04:35:10 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Herrera",
"Gonzalo",
""
],
[
"Huber",
"Patrick",
""
]
] |
2408.12022 | Lance Ying | Lance Ying, Tan Zhi-Xuan, Lionel Wong, Vikash Mansinghka, Joshua B.
Tenenbaum | Understanding Epistemic Language with a Language-augmented Bayesian
Theory of Mind | 23 pages; Published at the Transactions of the Association for
Computational Linguistics (TACL); Presented at NAACL 2025 | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | How do people understand and evaluate claims about others' beliefs, even
though these beliefs cannot be directly observed? In this paper, we introduce a
cognitive model of epistemic language interpretation, grounded in Bayesian
inferences about other agents' goals, beliefs, and intentions: a
language-augmented Bayesian theory-of-mind (LaBToM). By translating natural
language into an epistemic ``language-of-thought'' with grammar-constrained LLM
decoding, then evaluating these translations against the inferences produced by
inverting a generative model of rational action and perception, LaBToM captures
graded plausibility judgments of epistemic claims. We validate our model in an
experiment where participants watch an agent navigate a maze to find keys
hidden in boxes needed to reach their goal, then rate sentences about the
agent's beliefs. In contrast with multimodal LLMs (GPT-4o, Gemini Pro) and
ablated models, our model correlates highly with human judgments for a wide
range of expressions, including modal language, uncertainty expressions,
knowledge claims, likelihood comparisons, and attributions of false belief.
| [
{
"version": "v1",
"created": "Wed, 21 Aug 2024 22:29:56 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Apr 2025 15:31:32 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Ying",
"Lance",
""
],
[
"Zhi-Xuan",
"Tan",
""
],
[
"Wong",
"Lionel",
""
],
[
"Mansinghka",
"Vikash",
""
],
[
"Tenenbaum",
"Joshua B.",
""
]
] |
2408.13015 | Rui Li | Rui Li, Shikun Zhang, Zheng Qin, Chunxiao Du, Yang Zhou, and Zhisong
Xiao | High-dimentional Multipartite Entanglement Structure Detection with Low
Cost | null | null | 10.1103/PhysRevApplied.23.044033 | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantum entanglement detection and characterization are crucial for various
quantum information processes. Most existing methods for entanglement detection
rely heavily on a complete description of the quantum state, which requires
numerous measurements and complex setups. This makes these theoretically sound
approaches costly and impractical, as the system size increases. In this work,
we propose a multi-view neural network model to generate representations
suitable for entanglement structure detection. The number of required quantum
measurements is polynomial rather than exponential increase with the qubit
number. This remarkable reduction in resource costs makes it possible to detect
specific entanglement structures in large-scale systems. Numerical simulations
show that our method achieves over 95% detection accuracy for up to 19 qubits
systems. By enabling a universal, flexible and resource-efficient analysis of
entanglement structures, our approach enhances the capability of utilizing
quantum states across a wide range of applications.
| [
{
"version": "v1",
"created": "Fri, 23 Aug 2024 12:09:26 GMT"
}
] | 2025-04-21T00:00:00 | [
[
"Li",
"Rui",
""
],
[
"Zhang",
"Shikun",
""
],
[
"Qin",
"Zheng",
""
],
[
"Du",
"Chunxiao",
""
],
[
"Zhou",
"Yang",
""
],
[
"Xiao",
"Zhisong",
""
]
] |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.