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
2025-05-16 00:00:00
| authors_parsed
sequencelengths 1
2.83k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2505.10458 | Xulei Wang | Xulei Wang and Guohua Zhang | Smooth surface systems may contain smooth curves which have no measure
of maximal entropy | null | null | null | null | math.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we study Borel probability measures of maximal entropy for
analytic subsets in a dynamical system. It is well known that higher smoothness
of the map over smooth space plays important role in the study of invariant
measures of maximal entropy. A famous theorem of Newhouse states that smooth
diffeomorphisms on compact manifolds without boundary have invariant measures
of maximal entropy. However, we show that the situation becomes completely
different when we study measures of maximal entropy for analytic subsets.
Namely, we construct a smooth surface system which contains a smooth curve
having no Borel probability measure of maximal entropy. Another evidence to
show this difference is that, once an analytic set has one measure of maximal
entropy, then the set has many measures of maximal entropy (no matter if we
consider packing or Bowen entropy). For a general dynamical system with
positive entropy $h_\mathrm{top}(T)$, we shall show that the system contains
not only a Borel subset which has Borel probability measures of maximal entropy
and has entropy sufficiently close to $h_\mathrm{top}(T)$, but also a Borel
subset which has no Borel probability measures of maximal entropy and has
entropy equal to the arbitrarily given positive real number which is at most
$h_\mathrm{top}(T)$. We also provide in all $h$-expansive systems a full
characterization for analytic subsets which have Borel probability measures of
maximal entropy. Consequently, if let $Z\subset \mathbb{R}^n$ be any analytic
subset with positive Hausdorff dimension in Euclidean space, then the set $Z$
either has a measure of full lower Hausdorff dimension, or it can be
partitioned into a union of countably many analytic sets $\{Z_i\}_{i\in
\mathbb{N}}$ with $\dim_{\mathcal{H}}(Z_i) < \dim_{\mathcal{H}}(Z)$ for each
$i$.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:15:17 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Wang",
"Xulei",
""
],
[
"Zhang",
"Guohua",
""
]
] |
2505.10459 | Mitja Desmet | Mitja Desmet, Keito Watanabe, Tim Huege and Stijn Buitink | SMIET: Fast and accurate synthesis of radio pulses from extensive air
shower using simulated templates | 38 pages, 15 figures. Submitted to Astroparticle Physics | null | null | null | astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Interpreting the data from radio detectors for extensive air showers
typically relies on Monte-Carlo based simulation codes, which, despite their
accuracy are computationally expensive and present bottlenecks for analyses. To
address this issue we developed a novel method called template synthesis, which
synthesises the radio emission from cosmic ray air showers in seconds. This
hybrid approach uses a microscopically simulated, sliced shower (the origin) as
an input. It rescales the emission from each slice individually to synthesise
the emission from a shower with different properties (the target). In order to
be able to change the arrival direction during synthesis, we adjust the phases
based on the expected geometrical delays. We benchmark the method by comparing
synthesised traces to CoREAS simulations over a wide frequency range of [30,
500] MHz . The synthesis quality is primarily influenced by the difference in
$X_{max}$ between the origin and target shower. When $\Delta X_{max} \leq 100
g/cm^2$ , the scatter on the maximum amplitudes of the geomagnetic traces is at
most 4%. For the traces from the charge-excess component this scatter is
smaller than 6%. We also observe a bias with $\Delta X_{max}$ up to 5% for both
components, which appears to depend on the antenna position. Since the bias is
symmetrical around $\Delta X_{max} = 0 g/cm^2$, we can use an interpolation
approach to correct for it. We have implemented the template synthesis
algorithm in a Python package called \texttt{SMIET}, which includes all the
necessary parameters. This package has been successfully tested with air
showers with zenith angles up to $50^{\circ}$ and can be used with any
atmosphere, observation level and magnetic field. We demonstrate that the
synthesis quality remains comparable to our main benchmarks across various
scenarios and discuss use cases, including machine-learning-based analyses.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:16:15 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Desmet",
"Mitja",
""
],
[
"Watanabe",
"Keito",
""
],
[
"Huege",
"Tim",
""
],
[
"Buitink",
"Stijn",
""
]
] |
2505.10460 | Mukesh Kumar Vyas | Mukesh Kumar Vyas and Asaf Pe'er | Generation of magnetic fields around black hole accretion discs due to
non conservative radiation fields | 9 pages, 5 figures | null | null | null | astro-ph.HE physics.plasm-ph | http://creativecommons.org/licenses/by/4.0/ | We investigate the generation of magnetic fields above black hole accretion
discs due to the non-zero curl of the disc radiation field. By self
consistently computing the components of the radiation flux and their curl, we
show that the rotational nature of the radiation field induces charge
separation, leading to magnetic field generation in the plasma above the disc.
Solving the magnetohydrodynamic equations, we derive the time evolution of
these fields and demonstrate that they grow over astrophysically relevant
timescales. For a standard Keplerian accretion disc, the produced magnetic
fields remain weak, on the order of a few Gauss, consistent with previous
predictions. However, when a luminous corona is present in the inner disc
region ($r_d < 3-10 r_g$), the generated fields reach dynamically significant
strengths of up to $10^5$ Gauss where the magnetic energy density approaching
few percentage of equipartition with the gas pressure. These fields develop
within realistic growth timescales (such as viscous timescale) and can be
dynamically significant in governing disc and jet evolution. Our findings
suggest that radiation-driven magnetic fields play a crucial role in accretion
flow magnetization, influencing both disc dynamics and observational
signatures. The predicted field strengths could affect the thermal emission,
synchrotron radiation, and polarization properties of black hole accretion
systems, with implications for X-ray binaries, AGN, and jet formation. Future
numerical simulations and high-resolution polarimetric observations, such as
those from IXPE, eXTP, and EHT, may provide observational confirmation of our
findings.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:16:36 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Vyas",
"Mukesh Kumar",
""
],
[
"Pe'er",
"Asaf",
""
]
] |
2505.10461 | Uli Fahrenberg | Amazigh Amrane, Hugo Bazille, Emily Clement, Uli Fahrenberg, Marie
Fortin, Krzysztof Ziemia\'nski | B\"uchi-Elgot-Trakhtenbrot Theorem for Higher-Dimensional Automata | null | null | null | null | cs.FL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we explore languages of higher-dimensional automata (HDAs) from
an algebraic and logical point of view. Such languages are sets of finite
width-bounded interval pomsets with interfaces (ipomsets) closed under order
extension. We show that ipomsets can be represented as equivalence classes of
words over a particular alphabet, called step sequences. We introduce an
automaton model that recognize such languages. Doing so allows us to lift the
classical B\"uchi-Elgot-Trakhtenbrot Theorem to languages of HDAs: we prove
that a set of interval ipomsets is the language of an HDA if and only if it is
simultaneously MSO-definable, of bounded width, and closed under order
refinement.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:16:41 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Amrane",
"Amazigh",
""
],
[
"Bazille",
"Hugo",
""
],
[
"Clement",
"Emily",
""
],
[
"Fahrenberg",
"Uli",
""
],
[
"Fortin",
"Marie",
""
],
[
"Ziemiański",
"Krzysztof",
""
]
] |
2505.10462 | Giacomo Marmorini | Yoshiyuki Kakihara, Daisuke Yamamoto, Giacomo Marmorini | Compressed sensing quantum state tomography for qudits: A comparison of
Gell-Mann and Heisenberg-Weyl observable bases | 13 pages, 4 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Quantum state tomography (QST) is an essential technique for reconstructing
the density matrix of an unknown quantum state from measurement data, crucial
for quantum information processing. However, conventional QST requires an
exponentially growing number of measurements as the system dimension increases,
posing a significant challenge for high-dimensional systems. To mitigate this
issue, compressed sensing quantum state tomography (CS-QST) has been proposed,
significantly reducing the required number of measurements. In this study, we
investigate the impact of basis selection in CS-QST for qudit systems, which
are fundamental to high-dimensional quantum information processing.
Specifically, we compare the efficiency of the generalized Gell-Mann (GGM) and
Heisenberg-Weyl observable (HWO) bases by numerically reconstructing density
matrices and evaluating reconstruction accuracy using fidelity and trace
distance metrics. Our results demonstrate that, while both bases allow for
successful density matrix reconstruction, the HWO basis becomes more efficient
as the qudit dimension increases. Furthermore, we find the best fitting curves
that estimate the number of measurement operators required to achieve a
fidelity of at least 95%. These findings highlight the significance of basis
selection in CS-QST and provide valuable insights for optimizing measurement
strategies in high-dimensional quantum state tomography.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:17:26 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Kakihara",
"Yoshiyuki",
""
],
[
"Yamamoto",
"Daisuke",
""
],
[
"Marmorini",
"Giacomo",
""
]
] |
2505.10463 | Ye Li | Yi-Fang Liang, Ye Li, Zhen-Fan Tang, Xuan Yang, Song-Bo Zhang,
Yuan-Pei Yang, Fa-Yin Wang, Bao Wang, Di Xiao, Qing Zhao, Jun-Jie Wei,
Jin-Jun Geng, Jia-Rui Niu, Jun-Shuo Zhang, Guo Chen, Min Fang, Xue-Feng Wu,
Zi-Gao Dai, Wei-Wei Zhu, Peng Jiang, Bing Zhang | A possible periodic RM evolution in the repeating FRB 20220529 | 9 pages, 5 figures, 1 table | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Fast radio bursts (FRBs) are mysterious millisecond-duration radio transients
from the distant universe. Some of them repeat, while others do not. In order
to explore their origin, periodic examinations have been conducted on repeating
FRBs. Most of them show irregular properties, including burst rate, dispersion
measure (DM), and rotation measure (RM). A notable exception is FRB~20180916B,
which shows a significant 16-day periodic burst rate. Possible periodic
activities have also been reported in FRB~20121102A and FRB~20240209A. However,
periodic studies of other properties are sparse. FRB~20220529 was monitored by
the Five-hundred-meter Aperture Spherical radio Telescope (FAST) for nearly
three years, enabling periodic examinations of its properties. Here we report a
possible period of $\sim 200$ days in the RM evolution, with a significance of
4.2 $\sigma$ estimated via the Lomb-Scargle algorithm and 3.5 $\sigma$ with a
phase-folding method. The burst rate was also examined for periodicity. This is
consistent with the binary origin indicated by the significant RM increase and
its prompt recovery.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:17:57 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Liang",
"Yi-Fang",
""
],
[
"Li",
"Ye",
""
],
[
"Tang",
"Zhen-Fan",
""
],
[
"Yang",
"Xuan",
""
],
[
"Zhang",
"Song-Bo",
""
],
[
"Yang",
"Yuan-Pei",
""
],
[
"Wang",
"Fa-Yin",
""
],
[
"Wang",
"Bao",
""
],
[
"Xiao",
"Di",
""
],
[
"Zhao",
"Qing",
""
],
[
"Wei",
"Jun-Jie",
""
],
[
"Geng",
"Jin-Jun",
""
],
[
"Niu",
"Jia-Rui",
""
],
[
"Zhang",
"Jun-Shuo",
""
],
[
"Chen",
"Guo",
""
],
[
"Fang",
"Min",
""
],
[
"Wu",
"Xue-Feng",
""
],
[
"Dai",
"Zi-Gao",
""
],
[
"Zhu",
"Wei-Wei",
""
],
[
"Jiang",
"Peng",
""
],
[
"Zhang",
"Bing",
""
]
] |
2505.10464 | Jiaming Liang | Jiaming Liang, Lihuan Dai, Xiaoqi Sheng, Xiangguang Chen, Chun Yao,
Guihua Tao, Qibin Leng, Honming Cai, Xi Zhong | HWA-UNETR: Hierarchical Window Aggregate UNETR for 3D Multimodal Gastric
Lesion Segmentation | This work has been provisionally accepted for MICCAI 2025 | null | null | null | eess.IV cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multimodal medical image segmentation faces significant challenges in the
context of gastric cancer lesion analysis. This clinical context is defined by
the scarcity of independent multimodal datasets and the imperative to
amalgamate inherently misaligned modalities. As a result, algorithms are
constrained to train on approximate data and depend on application migration,
leading to substantial resource expenditure and a potential decline in analysis
accuracy. To address those challenges, we have made two major contributions:
First, we publicly disseminate the GCM 2025 dataset, which serves as the first
large-scale, open-source collection of gastric cancer multimodal MRI scans,
featuring professionally annotated FS-T2W, CE-T1W, and ADC images from 500
patients. Second, we introduce HWA-UNETR, a novel 3D segmentation framework
that employs an original HWA block with learnable window aggregation layers to
establish dynamic feature correspondences between different modalities'
anatomical structures, and leverages the innovative tri-orientated fusion mamba
mechanism for context modeling and capturing long-range spatial dependencies.
Extensive experiments on our GCM 2025 dataset and the publicly BraTS 2021
dataset validate the performance of our framework, demonstrating that the new
approach surpasses existing methods by up to 1.68\% in the Dice score while
maintaining solid robustness. The dataset and code are public via
https://github.com/JeMing-creater/HWA-UNETR.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:18:00 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Liang",
"Jiaming",
""
],
[
"Dai",
"Lihuan",
""
],
[
"Sheng",
"Xiaoqi",
""
],
[
"Chen",
"Xiangguang",
""
],
[
"Yao",
"Chun",
""
],
[
"Tao",
"Guihua",
""
],
[
"Leng",
"Qibin",
""
],
[
"Cai",
"Honming",
""
],
[
"Zhong",
"Xi",
""
]
] |
2505.10465 | Yizhou Liu | Yizhou liu, Ziming Liu, Jeff Gore | Superposition Yields Robust Neural Scaling | 30 pages, 23 figures | null | null | null | cs.LG cs.AI cs.CL | http://creativecommons.org/licenses/by/4.0/ | The success of today's large language models (LLMs) depends on the
observation that larger models perform better. However, the origin of this
neural scaling law -- the finding that loss decreases as a power law with model
size -- remains unclear. Starting from two empirical principles -- that LLMs
represent more things than the model dimensions (widths) they have (i.e.,
representations are superposed), and that words or concepts in language occur
with varying frequencies -- we constructed a toy model to study the loss
scaling with model size. We found that when superposition is weak, meaning only
the most frequent features are represented without interference, the scaling of
loss with model size depends on the underlying feature frequency; if feature
frequencies follow a power law, so does the loss. In contrast, under strong
superposition, where all features are represented but overlap with each other,
the loss becomes inversely proportional to the model dimension across a wide
range of feature frequency distributions. This robust scaling behavior is
explained geometrically: when many more vectors are packed into a lower
dimensional space, the interference (squared overlaps) between vectors scales
inversely with that dimension. We then analyzed four families of open-sourced
LLMs and found that they exhibit strong superposition and quantitatively match
the predictions of our toy model. The Chinchilla scaling law turned out to also
agree with our results. We conclude that representation superposition is an
important mechanism underlying the observed neural scaling laws. We anticipate
that these insights will inspire new training strategies and model
architectures to achieve better performance with less computation and fewer
parameters.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:18:13 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"liu",
"Yizhou",
""
],
[
"Liu",
"Ziming",
""
],
[
"Gore",
"Jeff",
""
]
] |
2505.10466 | Juehang Qin | Juehang Qin, Shixiao Liang, Christopher Tunnell | FlowVAT: Normalizing Flow Variational Inference with Affine-Invariant
Tempering | 10 pages, 5 figures, and 2 tables in main text, two appendices | null | null | null | stat.ML cs.LG stat.CO | http://creativecommons.org/licenses/by/4.0/ | Multi-modal and high-dimensional posteriors present significant challenges
for variational inference, causing mode-seeking behavior and collapse despite
the theoretical expressiveness of normalizing flows. Traditional annealing
methods require temperature schedules and hyperparameter tuning, falling short
of the goal of truly black-box variational inference. We introduce FlowVAT, a
conditional tempering approach for normalizing flow variational inference that
addresses these limitations. Our method tempers both the base and target
distributions simultaneously, maintaining affine-invariance under tempering. By
conditioning the normalizing flow on temperature, we leverage overparameterized
neural networks' generalization capabilities to train a single flow
representing the posterior across a range of temperatures. This preserves modes
identified at higher temperatures when sampling from the variational posterior
at $T = 1$, mitigating standard variational methods' mode-seeking behavior. In
experiments with 2, 10, and 20 dimensional multi-modal distributions, FlowVAT
outperforms traditional and adaptive annealing methods, finding more modes and
achieving better ELBO values, particularly in higher dimensions where existing
approaches fail. Our method requires minimal hyperparameter tuning and does not
require an annealing schedule, advancing toward fully-automatic black-box
variational inference for complicated posteriors.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:20:36 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Qin",
"Juehang",
""
],
[
"Liang",
"Shixiao",
""
],
[
"Tunnell",
"Christopher",
""
]
] |
2505.10467 | Pablo Hern\'andez-Garc\'ia | Pablo Hern\'andez-Garc\'ia, Daniel Hern\'andez Serrano, Dar\'io
S\'anchez G\'omez | From Persistence to Resilience: New Betti Numbers for Analyzing
Robustness in Simplicial Complex Networks | null | null | null | null | math.AT | http://creativecommons.org/licenses/by/4.0/ | Persistent homology is a fundamental tool in topological data analysis;
however, it lacks methods to quantify the fragility or fineness of cycles,
anticipate their formation or disappearance, or evaluate their stability beyond
persistence. Furthermore, classical Betti numbers fail to capture key
structural properties such as simplicial dimensions and higher-order
adjacencies. In this work, we investigate the robustness of simplicial networks
by analyzing cycle thickness and their resilience to failures or attacks. To
address these limitations, we draw inspiration from persistent homology to
introduce filtrations that model distinct simplicial elimination rules, leading
to the definition of two novel Betti number families: thick and cohesive Betti
numbers. These improved invariants capture richer structural information,
enabling the measurement of the thickness of the links in the homology cycle
and the assessment of the strength of their connections. This enhances and
refines classical topological descriptors and our approach provides deeper
insights into the structural dynamics of simplicial complexes and establishes a
theoretical framework for assessing robustness in higher-order networks.
Finally, we establish that the resilience of topological features to simplicial
attacks can be systematically examined through biparameter persistence modules,
wherein one parameter encodes the progression of the attack, and the other
captures structural refinements informed by thickness or cohesiveness.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:21:07 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Hernández-García",
"Pablo",
""
],
[
"Serrano",
"Daniel Hernández",
""
],
[
"Gómez",
"Darío Sánchez",
""
]
] |
2505.10468 | Konstantinos I. Roumeliotis | Ranjan Sapkota, Konstantinos I. Roumeliotis, Manoj Karkee | AI Agents vs. Agentic AI: A Conceptual Taxonomy, Applications and
Challenge | 32 pages, 14 figures, 11 tables | null | null | null | cs.AI | http://creativecommons.org/licenses/by/4.0/ | This study critically distinguishes between AI Agents and Agentic AI,
offering a structured conceptual taxonomy, application mapping, and challenge
analysis to clarify their divergent design philosophies and capabilities. We
begin by outlining the search strategy and foundational definitions,
characterizing AI Agents as modular systems driven by Large Language Models
(LLMs) and Large Image Models (LIMs) for narrow, task-specific automation.
Generative AI is positioned as a precursor, with AI Agents advancing through
tool integration, prompt engineering, and reasoning enhancements. In contrast,
Agentic AI systems represent a paradigmatic shift marked by multi-agent
collaboration, dynamic task decomposition, persistent memory, and orchestrated
autonomy. Through a sequential evaluation of architectural evolution,
operational mechanisms, interaction styles, and autonomy levels, we present a
comparative analysis across both paradigms. Application domains such as
customer support, scheduling, and data summarization are contrasted with
Agentic AI deployments in research automation, robotic coordination, and
medical decision support. We further examine unique challenges in each paradigm
including hallucination, brittleness, emergent behavior, and coordination
failure and propose targeted solutions such as ReAct loops, RAG, orchestration
layers, and causal modeling. This work aims to provide a definitive roadmap for
developing robust, scalable, and explainable AI agent and Agentic AI-driven
systems. >AI Agents, Agent-driven, Vision-Language-Models, Agentic AI Decision
Support System, Agentic-AI Applications
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:21:33 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Sapkota",
"Ranjan",
""
],
[
"Roumeliotis",
"Konstantinos I.",
""
],
[
"Karkee",
"Manoj",
""
]
] |
2505.10469 | Kohei Kawabata | Kohei Kawabata, Daichi Nakamura | Hopf Bifurcation of Nonlinear Non-Hermitian Skin Effect | 8+6 pages, 3+1 figures | null | null | null | cond-mat.mes-hall physics.optics quant-ph | http://creativecommons.org/licenses/by/4.0/ | The non-Hermitian skin effect, nonreciprocity-induced anomalous localization
of an extensive number of eigenstates, represents a hallmark of non-Hermitian
topological systems with no analogs in Hermitian systems. Despite its
significance across various open classical and quantum systems, the influence
of nonlinearity has remained largely unclear. Here, we reveal the Hopf
bifurcation of the nonlinear skin effect as a critical phenomenon unique to
nonlinear non-Hermitian systems. We demonstrate that nonlinearity destabilizes
skin states and instead gives rise to the emergence of delocalized states
associated with limit cycles in phase space. We also uncover the algebraically
localized critical skin effect precisely at the Hopf bifurcation point. We
illustrate these behavior in a nonlinear extension of the Hatano-Nelson model
in both continuum and lattice. Our work shows a significant role of
nonlinearity in the skin effect and uncovers rich phenomena arising from the
interplay between non-Hermiticity and nonlinearity.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:21:42 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Kawabata",
"Kohei",
""
],
[
"Nakamura",
"Daichi",
""
]
] |
2505.10470 | Olov Schavemaker | Olov Schavemaker | Separating balls with partly random hyperplanes with a view to partly
random neural networks | null | null | null | null | math.PR | http://creativecommons.org/licenses/by/4.0/ | We derive exact expressions for the probabilities that partly random
hyperplanes separate two Euclidean balls. The probability that a fully random
hyperplane separates two balls turns out to be significantly smaller than the
corresponding probabilities for hyperplanes which are not fully random in
certain cases. Our results motivate studying partially random neural networks
and provide a first step in this direction.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:22:27 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Schavemaker",
"Olov",
""
]
] |
2505.10471 | Qiuyang Mang | Jingbang Chen, Weinuo Li, Yingli Zhou, Hangrui Zhou, Qiuyang Mang, Can
Wang, Yixiang Fang, and Chenhao Ma | Scalable Approximate Biclique Counting over Large Bipartite Graphs | null | null | null | null | cs.SI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Counting $(p,q)$-bicliques in bipartite graphs is crucial for a variety of
applications, from recommendation systems to cohesive subgraph analysis. Yet,
it remains computationally challenging due to the combinatorial explosion to
exactly count the $(p,q)$-bicliques. In many scenarios, e.g., graph kernel
methods, however, exact counts are not strictly required. To design a scalable
and high-quality approximate solution, we novelly resort to $(p,q)$-broom, a
special spanning tree of the $(p,q)$-biclique, which can be counted via graph
coloring and efficient dynamic programming. Based on the intermediate results
of the dynamic programming, we propose an efficient sampling algorithm to
derive the approximate $(p,q)$-biclique count from the $(p,q)$-broom counts.
Theoretically, our method offers unbiased estimates with provable error
guarantees. Empirically, our solution outperforms existing approximation
techniques in both accuracy (up to 8$\times$ error reduction) and runtime (up
to 50$\times$ speedup) on nine real-world bipartite networks, providing a
scalable solution for large-scale $(p,q)$-biclique counting.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:22:37 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Chen",
"Jingbang",
""
],
[
"Li",
"Weinuo",
""
],
[
"Zhou",
"Yingli",
""
],
[
"Zhou",
"Hangrui",
""
],
[
"Mang",
"Qiuyang",
""
],
[
"Wang",
"Can",
""
],
[
"Fang",
"Yixiang",
""
],
[
"Ma",
"Chenhao",
""
]
] |
2505.10472 | Agnik Saha | Agnik Saha, Victoria Churchill, Anny D. Rodriguez, Ugur Kursuncu,
Muhammed Y. Idris | Large Language Models for Cancer Communication: Evaluating Linguistic
Quality, Safety, and Accessibility in Generative AI | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Effective communication about breast and cervical cancers remains a
persistent health challenge, with significant gaps in public understanding of
cancer prevention, screening, and treatment, potentially leading to delayed
diagnoses and inadequate treatments. This study evaluates the capabilities and
limitations of Large Language Models (LLMs) in generating accurate, safe, and
accessible cancer-related information to support patient understanding. We
evaluated five general-purpose and three medical LLMs using a mixed-methods
evaluation framework across linguistic quality, safety and trustworthiness, and
communication accessibility and affectiveness. Our approach utilized
quantitative metrics, qualitative expert ratings, and statistical analysis
using Welch's ANOVA, Games-Howell, and Hedges' g. Our results show that
general-purpose LLMs produced outputs of higher linguistic quality and
affectiveness, while medical LLMs demonstrate greater communication
accessibility. However, medical LLMs tend to exhibit higher levels of potential
harm, toxicity, and bias, reducing their performance in safety and
trustworthiness. Our findings indicate a duality between domain-specific
knowledge and safety in health communications. The results highlight the need
for intentional model design with targeted improvements, particularly in
mitigating harm and bias, and improving safety and affectiveness. This study
provides a comprehensive evaluation of LLMs for cancer communication, offering
critical insights for improving AI-generated health content and informing
future development of accurate, safe, and accessible digital health tools.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:23:21 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Saha",
"Agnik",
""
],
[
"Churchill",
"Victoria",
""
],
[
"Rodriguez",
"Anny D.",
""
],
[
"Kursuncu",
"Ugur",
""
],
[
"Idris",
"Muhammed Y.",
""
]
] |
2505.10473 | Fengdi Zhang | Fengdi Zhang, Hongkun Cao, Ruqi Huang | Consistent Quantity-Quality Control across Scenes for Deployment-Aware
Gaussian Splatting | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | To reduce storage and computational costs, 3D Gaussian splatting (3DGS) seeks
to minimize the number of Gaussians used while preserving high rendering
quality, introducing an inherent trade-off between Gaussian quantity and
rendering quality. Existing methods strive for better quantity-quality
performance, but lack the ability for users to intuitively adjust this
trade-off to suit practical needs such as model deployment under diverse
hardware and communication constraints. Here, we present ControlGS, a 3DGS
optimization method that achieves semantically meaningful and cross-scene
consistent quantity-quality control while maintaining strong quantity-quality
performance. Through a single training run using a fixed setup and a
user-specified hyperparameter reflecting quantity-quality preference, ControlGS
can automatically find desirable quantity-quality trade-off points across
diverse scenes, from compact objects to large outdoor scenes. It also
outperforms baselines by achieving higher rendering quality with fewer
Gaussians, and supports a broad adjustment range with stepless control over the
trade-off.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:23:51 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Zhang",
"Fengdi",
""
],
[
"Cao",
"Hongkun",
""
],
[
"Huang",
"Ruqi",
""
]
] |
2505.10474 | Kamil Kolincio | Kamil K. Kolincio, Marta Roman, Valerio Tammurello, Simone Di Cataldo,
Daniel Matulka, Sonia Francoual, Berthold St\"oger, Herwig Michor | Coexistence of charge density wave and field-tuned magnetic states in
TmNiC$_2$ | null | null | 10.1103/91vh-cpvx | null | cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Exploring the relations between coexisting, cooperative, or competing types
of ordering is a key to identify and harness the mechanisms governing the
mutual interactions between them, and to utilize their combined properties. We
have experimentally explored the response of the charge density wave (CDW) to
various antiferromagnetic, metamagnetic, and field-aligned ferromagnetic states
that constitute the magnetic phase diagram of TmNiC$_2$. The high resolution
x-ray diffraction experiment employing synchrotron radiation at low temperature
and high magnetic field, allowed to follow the superstructure satellite
reflections, being a sensitive probe of CDW. This investigation not only
reveals direct evidence that the charge density wave avoids even a partial
suppression in the antiferromagnetic ground state but also proves that this
state coexists, without any visible signatures of weakening, in the entire dome
of the magnetically ordered phases, including the field-aligned ferromagnetic
state. The calculations of the electronic and phonon structures support the
experiment, revealing that the dominant contribution to the CDW transition
stems from momentum-dependent electron-phonon coupling. We conclude that this
mechanism prevents the CDW from vanishing, although the nesting conditions
within the magnetically ordered phases deteriorate.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:24:02 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Kolincio",
"Kamil K.",
""
],
[
"Roman",
"Marta",
""
],
[
"Tammurello",
"Valerio",
""
],
[
"Di Cataldo",
"Simone",
""
],
[
"Matulka",
"Daniel",
""
],
[
"Francoual",
"Sonia",
""
],
[
"Stöger",
"Berthold",
""
],
[
"Michor",
"Herwig",
""
]
] |
2505.10475 | Mouxiang Chen | Mouxiang Chen, Binyuan Hui, Zeyu Cui, Jiaxi Yang, Dayiheng Liu,
Jianling Sun, Junyang Lin, Zhongxin Liu | Parallel Scaling Law for Language Models | null | null | null | null | cs.LG cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It is commonly believed that scaling language models should commit a
significant space or time cost, by increasing the parameters (parameter
scaling) or output tokens (inference-time scaling). We introduce the third and
more inference-efficient scaling paradigm: increasing the model's parallel
computation during both training and inference time. We apply $P$ diverse and
learnable transformations to the input, execute forward passes of the model in
parallel, and dynamically aggregate the $P$ outputs. This method, namely
parallel scaling (ParScale), scales parallel computation by reusing existing
parameters and can be applied to any model structure, optimization procedure,
data, or task. We theoretically propose a new scaling law and validate it
through large-scale pre-training, which shows that a model with $P$ parallel
streams is similar to scaling the parameters by $O(\log P)$ while showing
superior inference efficiency. For example, ParScale can use up to 22$\times$
less memory increase and 6$\times$ less latency increase compared to parameter
scaling that achieves the same performance improvement. It can also recycle an
off-the-shelf pre-trained model into a parallelly scaled one by post-training
on a small amount of tokens, further reducing the training budget. The new
scaling law we discovered potentially facilitates the deployment of more
powerful models in low-resource scenarios, and provides an alternative
perspective for the role of computation in machine learning.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:24:45 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Chen",
"Mouxiang",
""
],
[
"Hui",
"Binyuan",
""
],
[
"Cui",
"Zeyu",
""
],
[
"Yang",
"Jiaxi",
""
],
[
"Liu",
"Dayiheng",
""
],
[
"Sun",
"Jianling",
""
],
[
"Lin",
"Junyang",
""
],
[
"Liu",
"Zhongxin",
""
]
] |
2505.10476 | Urmi Ninad | Urmi Ninad, Jonas Wahl, Andreas Gerhardus, Jakob Runge | Causal discovery on vector-valued variables and consistency-guided
aggregation | null | null | null | null | stat.ME | http://creativecommons.org/licenses/by/4.0/ | Causal discovery (CD) aims to discover the causal graph underlying the data
generation mechanism of observed variables. In many real-world applications,
the observed variables are vector-valued, such as in climate science where
variables are defined over a spatial grid and the task is called
spatio-temporal causal discovery. We motivate CD in vector-valued variable
setting while considering different possibilities for the underlying model, and
highlight the pitfalls of commonly-used approaches when compared to a fully
vectorized approach. Furthermore, often the vector-valued variables are
high-dimensional, and aggregations of the variables, such as averages, are
considered in interest of efficiency and robustness. In the absence of
interventional data, testing for the soundness of aggregate variables as
consistent abstractions that map a low-level to a high-level structural causal
model (SCM) is hard, and recent works have illustrated the stringency of
conditions required for testing consistency. In this work, we take a careful
look at the task of vector-valued CD via constraint-based methods, focusing on
the problem of consistency of aggregation for this task. We derive three
aggregation consistency scores, based on compatibility of independence models
and (partial) aggregation, that quantify different aspects of the soundness of
an aggregation map for the CD problem. We present the argument that the
consistency of causal abstractions must be separated from the task-dependent
consistency of aggregation maps. As an actionable conclusion of our findings,
we propose a wrapper Adag to optimize a chosen aggregation consistency score
for aggregate-CD, to make the output of CD over aggregate variables more
reliable. We supplement all our findings with experimental evaluations on
synthetic non-time series and spatio-temporal data.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:24:46 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Ninad",
"Urmi",
""
],
[
"Wahl",
"Jonas",
""
],
[
"Gerhardus",
"Andreas",
""
],
[
"Runge",
"Jakob",
""
]
] |
2505.10477 | Gaurav Malik | Gaurav Rudra Malik, Rohit Kumar Shukla, S. Aravinda and Sunil Kumar
Mishra | Characterizing quantum dynamics using multipartite entanglement
generation | 5 pages, 3 figures. Accepted to and published in the Proceedings of
the 2025 17th International Conference on COMmunication Systems and NETworks
(COMSNETS), IEEE, pp. 1051-1055, Jan. 2025 | null | 10.1109/COMSNETS63942.2025.10885762 | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Entanglement is a defining feature of many-body quantum systems and is an
essential requirement for quantum computing. It is therefore useful to study
physical processes which generate entanglement within a large system, as they
maybe replicated for applications involving the said requirements in quantum
information processing. A possible avenue to maximize entanglement generation
is to rely on the phenomena of information scrambling, i.e. transport of
initially localized information throughout the system. Here the rationale is
that the spread of information carries with it an inherent capacity of
entanglement generation. Scrambling greatly depends upon the dynamical nature
of the system Hamiltonian, and the interplay between entanglement generation
and information scrambling maybe investigated taking a chain of interacting
spins on a one dimensional lattice. This system is analogous to an array of
qubits and this relative simplicity implies that the resulting unitary dynamics
can be efficiently simulated using present-day cloud based NISQ devices. In our
present work, we consider such a spin model which is made up of nearest and
next nearest neighbor XXZ Model, along with an introduced coupling term lambda.
This coupling term serves as a tuning parameter which modifies the dynamical
nature of the system from the integrable to the quantum chaotic regime. In
order to quantify the entanglement generated within the system we use the more
general multipartite metric which computes the average entanglement across all
system bipartitions to obtain a global picture of the entanglement structure
within the entire system.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:25:03 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Malik",
"Gaurav Rudra",
""
],
[
"Shukla",
"Rohit Kumar",
""
],
[
"Aravinda",
"S.",
""
],
[
"Mishra",
"Sunil Kumar",
""
]
] |
2505.10478 | Wei-Min Gu | Hao-Bin Liu, Wei-Min Gu, Yongqi Lu, Teng Liu, Jin-Zhong Liu | Searching for Accreting Compact Binaries in eROSITA DR1 | 10 pages, 4 figures, to be submitted | null | null | null | astro-ph.HE astro-ph.GA astro-ph.SR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Compact binaries with accreting white dwarfs, neutron stars, or black holes
are crucial for understanding accretion physics. In this study, we identify
accreting compact binary candidates in the eROSITA DR1 by combining Gaia
astrometry and ZTF time-domain photometry. We select accreting compact binary
candidates whose X-ray emission is likely accretion-driven, based on their
X-ray-to-optical flux ratios and optical colors, and further identify systems
with short-period ZTF light curves, thereby ruling out AGNs. The resulting 22
candidates, including two known compact binaries, represent promising
candidates for spectroscopic follow-up to confirm their accreting nature and
explore their physical properties. Our results demonstrate the effectiveness of
combining X-ray-to-optical flux ratio and optical color, and time-domain
photometry in selecting accreting compact binaries. The method is scalable and
adaptable to future multi-wavelength sky surveys, offering a promising path
toward a more complete census of compact binaries in the Galaxy.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:25:16 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Liu",
"Hao-Bin",
""
],
[
"Gu",
"Wei-Min",
""
],
[
"Lu",
"Yongqi",
""
],
[
"Liu",
"Teng",
""
],
[
"Liu",
"Jin-Zhong",
""
]
] |
2505.10479 | Kelly Victor-French | Kelly Victor-French, Karl Battams, and Brian E. Wood | The Relationship between the Kinematics of Coronal Mass Ejections and
the Brightness of the Corona | 18 pages, 15 figures, accepted for publication in The Astrophysical
Journal | null | 10.3847/1538-4357/add3fb | null | astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | We present an investigation into an apparent relationship between white-light
coronal brightness and the kinematics of flare-associated CMEs. Using a unique
dataset known as the LASCO Coronal Brightness Index (CBI), we conduct a study
that explores the brightness in the lower solar corona and its relationship to
the velocity of flare-associated coronal mass ejections (CMEs). We analyze all
M- and X-class flares that take place on or near the limbs of the Earth-facing
disk of the Sun between 1996 and 2022, determine if these flares are associated
with CMEs, and record the projection-corrected velocity of the eruptions if
they occurred. Using the CBI dataset, we evaluate the brightness in the corona
directly overlying the flare source locations between 2.4 and 6.2 solar radii,
and find that above a certain level of coronal brightness, the likelihood of a
high-velocity CME significantly decreases. This result implies coronal
brightness could be an important indicator of the kinematics of solar CMEs. We
also highlight and discuss the unique nature of Active Region (AR) 12192 in
2014, observing that its unprecedented overlying coronal brightness may be
related to the low CME productivity of that region.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:28:23 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Victor-French",
"Kelly",
""
],
[
"Battams",
"Karl",
""
],
[
"Wood",
"Brian E.",
""
]
] |
2505.10480 | Patricia Klein | Ilani Axelrod-Freed, Hanson Hao, Matthew Kendall, Patricia Klein, and
Patricia Klein | Some algebraic properties of ASM varieties | From UMN REU, comments welcome | null | null | null | math.CO math.AC | http://creativecommons.org/licenses/by/4.0/ | Fulton's matrix Schubert varieties are affine varieties that arise in the
study of Schubert calculus in the complete flag variety. Weigandt showed that
arbitrary intersections of matrix Schubert varieties, now called ASM varieties,
are indexed by alternating sign matrices (ASMs), objects with a long history in
enumerative combinatorics. It is very difficult to assess Cohen-Macaulayness of
ASM varieties or to compute their codimension, though these properties are well
understood for matrix Schubert varieties due to work of Fulton. In this paper
we study these properties of ASM varieties with a focus on the relationship
between a pair of ASMs and their direct sum. We also consider ASM pattern
avoidance from an algebro-geometric perspective.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:28:50 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Axelrod-Freed",
"Ilani",
""
],
[
"Hao",
"Hanson",
""
],
[
"Kendall",
"Matthew",
""
],
[
"Klein",
"Patricia",
""
],
[
"Klein",
"Patricia",
""
]
] |
2505.10481 | Ilya Ovodov | Ilya Ovodov, Petr Surovtsev, Karina Kvanchiani, Alexander Kapitanov,
Alexander Nagaev | Logos as a Well-Tempered Pre-train for Sign Language Recognition | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | This paper examines two aspects of the isolated sign language recognition
(ISLR) task. First, despite the availability of a number of datasets, the
amount of data for most individual sign languages is limited. It poses the
challenge of cross-language ISLR model training, including transfer learning.
Second, similar signs can have different semantic meanings. It leads to
ambiguity in dataset labeling and raises the question of the best policy for
annotating such signs. To address these issues, this study presents Logos, a
novel Russian Sign Language (RSL) dataset, the most extensive ISLR dataset by
the number of signers and one of the largest available datasets while also the
largest RSL dataset in size and vocabulary. It is shown that a model,
pre-trained on the Logos dataset can be used as a universal encoder for other
language SLR tasks, including few-shot learning. We explore cross-language
transfer learning approaches and find that joint training using multiple
classification heads benefits accuracy for the target lowresource datasets the
most. The key feature of the Logos dataset is explicitly annotated visually
similar sign groups. We show that explicitly labeling visually similar signs
improves trained model quality as a visual encoder for downstream tasks. Based
on the proposed contributions, we outperform current state-of-the-art results
for the WLASL dataset and get competitive results for the AUTSL dataset, with a
single stream model processing solely RGB video. The source code, dataset, and
pre-trained models are publicly available.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:31:49 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Ovodov",
"Ilya",
""
],
[
"Surovtsev",
"Petr",
""
],
[
"Kvanchiani",
"Karina",
""
],
[
"Kapitanov",
"Alexander",
""
],
[
"Nagaev",
"Alexander",
""
]
] |
2505.10482 | Ningyuan Yang | Ningyuan Yang, Jiaxuan Gao, Feng Gao, Yi Wu, Chao Yu | Fine-tuning Diffusion Policies with Backpropagation Through Diffusion
Timesteps | 9 pages for main text, 23 pages in total, submitted to Neurips, 13
figures | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Diffusion policies, widely adopted in decision-making scenarios such as
robotics, gaming and autonomous driving, are capable of learning diverse skills
from demonstration data due to their high representation power. However, the
sub-optimal and limited coverage of demonstration data could lead to diffusion
policies that generate sub-optimal trajectories and even catastrophic failures.
While reinforcement learning (RL)-based fine-tuning has emerged as a promising
solution to address these limitations, existing approaches struggle to
effectively adapt Proximal Policy Optimization (PPO) to diffusion models. This
challenge stems from the computational intractability of action likelihood
estimation during the denoising process, which leads to complicated
optimization objectives. In our experiments starting from randomly initialized
policies, we find that online tuning of Diffusion Policies demonstrates much
lower sample efficiency compared to directly applying PPO on MLP policies
(MLP+PPO). To address these challenges, we introduce NCDPO, a novel framework
that reformulates Diffusion Policy as a noise-conditioned deterministic policy.
By treating each denoising step as a differentiable transformation conditioned
on pre-sampled noise, NCDPO enables tractable likelihood evaluation and
gradient backpropagation through all diffusion timesteps. Our experiments
demonstrate that NCDPO achieves sample efficiency comparable to MLP+PPO when
training from scratch, outperforming existing methods in both sample efficiency
and final performance across diverse benchmarks, including continuous robot
control and multi-agent game scenarios. Furthermore, our experimental results
show that our method is robust to the number denoising timesteps in the
Diffusion Policy.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:33:44 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Yang",
"Ningyuan",
""
],
[
"Gao",
"Jiaxuan",
""
],
[
"Gao",
"Feng",
""
],
[
"Wu",
"Yi",
""
],
[
"Yu",
"Chao",
""
]
] |
2505.10483 | Yi Li | Yi Li, Haonan Wang, Qixiang Zhang, Boyu Xiao, Chenchang Hu, Hualiang
Wang, Xiaomeng Li | UniEval: Unified Holistic Evaluation for Unified Multimodal
Understanding and Generation | UniEval is the first evaluation framework designed for unified
multimodal models, including a holistic benchmark UniBench and the UniScore
metric | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | The emergence of unified multimodal understanding and generation models is
rapidly attracting attention because of their ability to enhance
instruction-following capabilities while minimizing model redundancy. However,
there is a lack of a unified evaluation framework for these models, which would
enable an elegant, simplified, and overall evaluation. Current models conduct
evaluations on multiple task-specific benchmarks, but there are significant
limitations, such as the lack of overall results, errors from extra evaluation
models, reliance on extensive labeled images, benchmarks that lack diversity,
and metrics with limited capacity for instruction-following evaluation. To
tackle these challenges, we introduce UniEval, the first evaluation framework
designed for unified multimodal models without extra models, images, or
annotations. This facilitates a simplified and unified evaluation process. The
UniEval framework contains a holistic benchmark, UniBench (supports both
unified and visual generation models), along with the corresponding UniScore
metric. UniBench includes 81 fine-grained tags contributing to high diversity.
Experimental results indicate that UniBench is more challenging than existing
benchmarks, and UniScore aligns closely with human evaluations, surpassing
current metrics. Moreover, we extensively evaluated SoTA unified and visual
generation models, uncovering new insights into Univeral's unique values.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:34:50 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Li",
"Yi",
""
],
[
"Wang",
"Haonan",
""
],
[
"Zhang",
"Qixiang",
""
],
[
"Xiao",
"Boyu",
""
],
[
"Hu",
"Chenchang",
""
],
[
"Wang",
"Hualiang",
""
],
[
"Li",
"Xiaomeng",
""
]
] |
2505.10484 | Andrea Baisero | Andrea Baisero and Rupali Bhati and Shuo Liu and Aathira Pillai and
Christopher Amato | Fixing Incomplete Value Function Decomposition for Multi-Agent
Reinforcement Learning | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Value function decomposition methods for cooperative multi-agent
reinforcement learning compose joint values from individual per-agent
utilities, and train them using a joint objective. To ensure that the action
selection process between individual utilities and joint values remains
consistent, it is imperative for the composition to satisfy the
individual-global max (IGM) property. Although satisfying IGM itself is
straightforward, most existing methods (e.g., VDN, QMIX) have limited
representation capabilities and are unable to represent the full class of IGM
values, and the one exception that has no such limitation (QPLEX) is
unnecessarily complex. In this work, we present a simple formulation of the
full class of IGM values that naturally leads to the derivation of QFIX, a
novel family of value function decomposition models that expand the
representation capabilities of prior models by means of a thin "fixing" layer.
We derive multiple variants of QFIX, and implement three variants in two
well-known multi-agent frameworks. We perform an empirical evaluation on
multiple SMACv2 and Overcooked environments, which confirms that QFIX (i)
succeeds in enhancing the performance of prior methods, (ii) learns more stably
and performs better than its main competitor QPLEX, and (iii) achieves this
while employing the simplest and smallest mixing models.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:36:18 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Baisero",
"Andrea",
""
],
[
"Bhati",
"Rupali",
""
],
[
"Liu",
"Shuo",
""
],
[
"Pillai",
"Aathira",
""
],
[
"Amato",
"Christopher",
""
]
] |
2505.10485 | Ali Ghorashi | Ali Ghorashi, Sachin Vaidya, Ziming Liu, Charlotte Loh, Thomas
Christensen, Max Tegmark, and Marin Solja\v{c}i\'c | Interpretable Artificial Intelligence for Topological Photonics | null | null | null | null | physics.optics | http://creativecommons.org/licenses/by/4.0/ | Topological photonic crystals (PhCs) offer robust, disorder-resistant modes
engendered by nontrivial band symmetries, but designing PhCs with prescribed
topological band properties remains a challenge due to the complexity involved
in mapping the continuous real-space design landscape of photonic crystals to
the discrete output space of band topology. Here, we introduce a new approach
to address this problem, employing Kolmogorov--Arnold networks (KANs) to
predict and inversely design the band symmetries of two-dimensional PhCs with
two-fold rotational (C2) symmetry. We show that a single-hidden-layer KAN,
trained on a dataset of C2-symmetric unit cells, achieves 99% accuracy in
classifying the symmetry eigenvalues of the lowest transverse-magnetic band.
For interpretability, we use symbolic regression to extract eight algebraic
formulas that characterize the band symmetry classes in terms of the Fourier
components of the dielectric function. These formulas not only retain the full
predictive power of the network but also enable deterministic inverse design.
Applying them, we generate 2,000 photonic crystals with target band symmetries,
achieving over 92% accuracy even for high-contrast, experimentally realizable
structures beyond the training domain.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:38:19 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Ghorashi",
"Ali",
""
],
[
"Vaidya",
"Sachin",
""
],
[
"Liu",
"Ziming",
""
],
[
"Loh",
"Charlotte",
""
],
[
"Christensen",
"Thomas",
""
],
[
"Tegmark",
"Max",
""
],
[
"Soljačić",
"Marin",
""
]
] |
2505.10486 | Julien Fageot | Julien Fageot | Variational Seasonal-Trend Decomposition with Sparse Continuous-Domain
Regularization | null | null | null | null | math.FA | http://creativecommons.org/licenses/by/4.0/ | We consider the inverse problem of recovering a continuous-domain function
from a finite number of noisy linear measurements. The unknown signal is
modeled as the sum of a slowly varying trend and a periodic or quasi-periodic
seasonal component. We formulate a variational framework for their joint
recovery by introducing convex regularizations based on generalized total
variation, which promote sparsity in spline-like representations. Our analysis
is conducted in an infinite-dimensional setting and leads to a representer
theorem showing that minimizers are splines in both components. To make the
approach numerically feasible, we introduce a family of discrete approximations
and prove their convergence to the original problem in the sense of
$\Gamma$-convergence. This further ensures the uniform convergence of discrete
solutions to their continuous counterparts. The proposed framework offers a
principled approach to seasonal-trend decomposition in the presence of noise
and limited measurements, with theoretical guarantees on both representation
and discretization.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:42:14 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Fageot",
"Julien",
""
]
] |
2505.10487 | Vincenzo Mariani | Vincenzo Mariani, Agn\`es Fienga, Zachary Murray, Micka\"el Gastineau
and Jacques Laskar | Boosting decision trees for Main Belt Asteroid selection in planetary
ephemerides: an alternative model | 16 pages: 11 pages for main body, 5 pages for appendix. 11 figures | null | null | null | astro-ph.EP astro-ph.IM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | One of the main bottleneck in assessing the accuracy of Mars orbit is the
unknown value of the asteroids in the Main Asteroid Belt. Nowadays a modeling
with 343 asteroids as point masses is used, with the relative masses fitted to
observational data. In the current work we propose an innovative methodology to
reduce the number of asteroids implemented as point masses, thus reducing the
number of parameters to be fitted, without a significant degradation of the
postfit residuals.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:42:53 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Mariani",
"Vincenzo",
""
],
[
"Fienga",
"Agnès",
""
],
[
"Murray",
"Zachary",
""
],
[
"Gastineau",
"Mickaël",
""
],
[
"Laskar",
"Jacques",
""
]
] |
2505.10488 | Tomoya Suzuguchi | Tomoya Suzuguchi, Hidetoshi Omiya, Hiroki Takeda | Possibility of Multi-Messenger Observations of Quasi-Periodic Eruptions
with X-rays and Gravitational Waves | 11 pages, 7 figures. Comments welcome!!! | null | null | null | astro-ph.HE gr-qc | http://creativecommons.org/licenses/by/4.0/ | Recent X-ray observations have discovered a class of periodic X-ray flares in
galactic nuclei known as quasi-periodic eruptions (QPEs). A promising
explanation of QPEs is an emission produced when a stellar-mass object orbiting
a central supermassive black hole crosses an accretion disk. If the companion
is a compact object, such systems would be a prospective multi-messenger target
for the space-based observatory LISA and its successors. Here we quantify the
prospects for joint X-ray and GW detection of QPEs with orbital frequency in
the mHz band using a minimal flare-emission model. Our analysis shows that
X-ray observations are most effective at orbital frequencies up to roughly 1
mHz, whereas LISA is sensitive chiefly above about 1 mHz. Because the optimal
sensitivity windows overlap only marginally, we predict at most one joint
detection during LISA's nominal mission lifetime. Extending GW sensitivity into
the sub-millihertz regime (< 0.1 mHz) would raise the possibility of the joint
detection by an order of magnitude, enabling QPEs as an interesting
multi-messenger target.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:43:05 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Suzuguchi",
"Tomoya",
""
],
[
"Omiya",
"Hidetoshi",
""
],
[
"Takeda",
"Hiroki",
""
]
] |
2505.10489 | Ilya Deriy | Ilya Deriy, Danil Kornovan, Mihail Petrov, and Andrey Bogdanov | Optical Spintronics: Towards Lorentz-Reciprocal Nonreciprocal Photonics | null | null | null | null | physics.optics | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Energy, momentum, and angular momentum are fundamental properties tied to the
symmetries of space and time, with photons and other elementary particles
acting as carriers of these quantities. In most optical and optoelectronic
devices, energy transfer is crucial, but it often results in undesirable energy
absorption. Moreover, non-reciprocal elements such as optical diodes and
circulators are difficult to implement in photonics, as they typically require
time-dependent perturbations, nonlinear effects, or external magnetic fields.
This presents a significant barrier to the development of efficient, compact
photonic technologies. We introduce the concept of optical spin current,
wherein spin angular momentum is transferred by an electromagnetic field
without accompanying energy transfer. This phenomenon is analogous to electron
spin currents, where spin is decoupled from charge flow. Building on this
principle, we propose optical spin diode and circulator -- devices that enable
unidirectional propagation of spin currents while maintaining bidirectional
energy flow, thus preserving reciprocity. Furthermore, we demonstrate
asymmetric spin transfer between quantum dots mediated by the optical spin
diode, highlighting the potential for novel optical spintronic functionalities.
These findings lay the foundation for devices that leverage optical spin
transfer, opening new avenues for advancements in optical spintronics.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:43:52 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Deriy",
"Ilya",
""
],
[
"Kornovan",
"Danil",
""
],
[
"Petrov",
"Mihail",
""
],
[
"Bogdanov",
"Andrey",
""
]
] |
2505.10490 | Steffen Becker | Leon Hannig and Annika Bush and Meltem Aksoy and Steffen Becker and
Greta Ontrup | Campus AI vs Commercial AI: A Late-Breaking Study on How LLM
As-A-Service Customizations Shape Trust and Usage Patterns | null | null | null | null | cs.CY | http://creativecommons.org/licenses/by/4.0/ | As the use of Large Language Models (LLMs) by students, lecturers and
researchers becomes more prevalent, universities - like other organizations -
are pressed to develop coherent AI strategies. LLMs as-a-Service (LLMaaS) offer
accessible pre-trained models, customizable to specific (business) needs. While
most studies prioritize data, model, or infrastructure adaptations (e.g., model
fine-tuning), we focus on user-salient customizations, like interface changes
and corporate branding, which we argue influence users' trust and usage
patterns. This study serves as a functional prequel to a large-scale field
study in which we examine how students and employees at a German university
perceive and use their institution's customized LLMaaS compared to ChatGPT. The
goals of this prequel are to stimulate discussions on psychological effects of
LLMaaS customizations and refine our research approach through feedback. Our
forthcoming findings will deepen the understanding of trust dynamics in LLMs,
providing practical guidance for organizations considering LLMaaS deployment.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:45:33 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Hannig",
"Leon",
""
],
[
"Bush",
"Annika",
""
],
[
"Aksoy",
"Meltem",
""
],
[
"Becker",
"Steffen",
""
],
[
"Ontrup",
"Greta",
""
]
] |
2505.10491 | Sonali Patnaik Dr. | Sonali Patnaik, Lopamudra Nayak and Sanjay Kumar Swain | Phenomenological study of semitauonic decays of $B$ and $B_s$ mesons to
charm states | 13 pages, 5 figures. Feedbacks are welcome | null | null | null | hep-ph | http://creativecommons.org/licenses/by/4.0/ | In this work, we compute the branching fraction of $B \to D^{(*)} \,\tau \,
\nu_{\tau}$ and $B_s \to D_s^{(*)}\, \tau\, \nu_{\tau}$ within the Relativistic
Independent Quark Model, emphasizing the harmonic potential model-dependent
analysis of these decay channels in the precision flavor physics era.
Considering the experimental observation of longitudinal $\tau$-polarization
and fraction of longitudinal polarization ($F_L$) at LHCb and Belle, we have
also investigated these observables within our model framework which are
aligning well with the standard model expectations. We perform a comprehensive
analysis of the form factors across the whole accessible kinematic range of
$q^2$. Our results are consistent and agree well with other theoretical
approaches as well as with the experimental measurements. Furthermore, we
evaluated the clean ratios of $B_s$ to $B_0$ in the semimuonic mode that are in
accordance with the LHCb measurements, and support the validity of the SU(3)
flavor symmetry. Although the concept of new physics is facing a significant
challenge at the current TeV scale, semileptonic $B$ decays, nevertheless,
always serve as a valuable probe to study the decay dynamics.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:46:50 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Patnaik",
"Sonali",
""
],
[
"Nayak",
"Lopamudra",
""
],
[
"Swain",
"Sanjay Kumar",
""
]
] |
2505.10492 | Taylor Bobrow | Taylor L. Bobrow, Mayank Golhar, Suchapa Arayakarnkul, Anthony A.
Song, Saowanee Ngamruengphong, Nicholas J. Durr | Multi-contrast laser endoscopy for in vivo gastrointestinal imaging | null | null | null | null | eess.IV cs.CV physics.med-ph physics.optics | http://creativecommons.org/licenses/by/4.0/ | White light endoscopy is the clinical gold standard for detecting diseases in
the gastrointestinal tract. Most applications involve identifying visual
abnormalities in tissue color, texture, and shape. Unfortunately, the contrast
of these features is often subtle, causing many clinically relevant cases to go
undetected. To overcome this challenge, we introduce Multi-contrast Laser
Endoscopy (MLE): a platform for widefield clinical imaging with rapidly tunable
spectral, coherent, and directional illumination. We demonstrate three
capabilities of MLE: enhancing tissue chromophore contrast with multispectral
diffuse reflectance, quantifying blood flow using laser speckle contrast
imaging, and characterizing mucosal topography using photometric stereo. We
validate MLE with benchtop models, then demonstrate MLE in vivo during clinical
colonoscopies. MLE images from 31 polyps demonstrate an approximate three-fold
improvement in contrast and a five-fold improvement in color difference
compared to white light and narrow band imaging. With the ability to reveal
multiple complementary types of tissue contrast while seamlessly integrating
into the clinical environment, MLE shows promise as an investigative tool to
improve gastrointestinal imaging.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:47:24 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Bobrow",
"Taylor L.",
""
],
[
"Golhar",
"Mayank",
""
],
[
"Arayakarnkul",
"Suchapa",
""
],
[
"Song",
"Anthony A.",
""
],
[
"Ngamruengphong",
"Saowanee",
""
],
[
"Durr",
"Nicholas J.",
""
]
] |
2505.10493 | Shaohan Wang | Shaohan Wang, Licheng Zhang, Zheren Fu, Zhendong Mao | CL-RAG: Bridging the Gap in Retrieval-Augmented Generation with
Curriculum Learning | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Retrieval-Augmented Generation (RAG) is an effective method to enhance the
capabilities of large language models (LLMs). Existing methods focus on
optimizing the retriever or generator in the RAG system by directly utilizing
the top-k retrieved documents. However, the documents effectiveness are various
significantly across user queries, i.e. some documents provide valuable
knowledge while others totally lack critical information. It hinders the
retriever and generator's adaptation during training. Inspired by human
cognitive learning, curriculum learning trains models using samples progressing
from easy to difficult, thus enhancing their generalization ability, and we
integrate this effective paradigm to the training of the RAG system. In this
paper, we propose a multi-stage Curriculum Learning based RAG system training
framework, named CL-RAG. We first construct training data with multiple
difficulty levels for the retriever and generator separately through sample
evolution. Then, we train the model in stages based on the curriculum learning
approach, thereby optimizing the overall performance and generalization of the
RAG system more effectively. Our CL-RAG framework demonstrates consistent
effectiveness across four open-domain QA datasets, achieving performance gains
of 2% to 4% over multiple advanced methods.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:53:04 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Wang",
"Shaohan",
""
],
[
"Zhang",
"Licheng",
""
],
[
"Fu",
"Zheren",
""
],
[
"Mao",
"Zhendong",
""
]
] |
2505.10494 | Yutao Mou | Yutao Mou, Xiao Deng, Yuxiao Luo, Shikun Zhang, Wei Ye | Can You Really Trust Code Copilots? Evaluating Large Language Models
from a Code Security Perspective | Accepted by ACL2025 Main Conference | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Code security and usability are both essential for various coding assistant
applications driven by large language models (LLMs). Current code security
benchmarks focus solely on single evaluation task and paradigm, such as code
completion and generation, lacking comprehensive assessment across dimensions
like secure code generation, vulnerability repair and discrimination. In this
paper, we first propose CoV-Eval, a multi-task benchmark covering various tasks
such as code completion, vulnerability repair, vulnerability detection and
classification, for comprehensive evaluation of LLM code security. Besides, we
developed VC-Judge, an improved judgment model that aligns closely with human
experts and can review LLM-generated programs for vulnerabilities in a more
efficient and reliable way. We conduct a comprehensive evaluation of 20
proprietary and open-source LLMs. Overall, while most LLMs identify vulnerable
codes well, they still tend to generate insecure codes and struggle with
recognizing specific vulnerability types and performing repairs. Extensive
experiments and qualitative analyses reveal key challenges and optimization
directions, offering insights for future research in LLM code security.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:53:41 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Mou",
"Yutao",
""
],
[
"Deng",
"Xiao",
""
],
[
"Luo",
"Yuxiao",
""
],
[
"Zhang",
"Shikun",
""
],
[
"Ye",
"Wei",
""
]
] |
2505.10495 | Dewang Sultania | Vibha Belavadi, Tushar Vatsa, Dewang Sultania, Suhas Suresha, Ishita
Verma, Cheng Chen, Tracy Holloway King, Michael Friedrich | RouteNator: A Router-Based Multi-Modal Architecture for Generating
Synthetic Training Data for Function Calling LLMs | Proceedings of the 4th International Workshop on Knowledge-Augmented
Methods for Natural Language Processing | https://aclanthology.org/2025.knowledgenlp-1.10/ KnowledgeNLP 2025 | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | This paper addresses fine-tuning Large Language Models (LLMs) for function
calling tasks when real user interaction data is unavailable. In digital
content creation tools, where users express their needs through natural
language queries that must be mapped to API calls, the lack of real-world
task-specific data and privacy constraints for training on it necessitate
synthetic data generation. Existing approaches to synthetic data generation
fall short in diversity and complexity, failing to replicate real-world data
distributions and leading to suboptimal performance after LLM fine-tuning. We
present a novel router-based architecture that leverages domain resources like
content metadata and structured knowledge graphs, along with text-to-text and
vision-to-text language models to generate high-quality synthetic training
data. Our architecture's flexible routing mechanism enables synthetic data
generation that matches observed real-world distributions, addressing a
fundamental limitation of traditional approaches. Evaluation on a comprehensive
set of real user queries demonstrates significant improvements in both function
classification accuracy and API parameter selection. Models fine-tuned with our
synthetic data consistently outperform traditional approaches, establishing new
benchmarks for function calling tasks.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:53:45 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Belavadi",
"Vibha",
""
],
[
"Vatsa",
"Tushar",
""
],
[
"Sultania",
"Dewang",
""
],
[
"Suresha",
"Suhas",
""
],
[
"Verma",
"Ishita",
""
],
[
"Chen",
"Cheng",
""
],
[
"King",
"Tracy Holloway",
""
],
[
"Friedrich",
"Michael",
""
]
] |
2505.10496 | Raman Dutt | Raman Dutt, Pedro Sanchez, Yongchen Yao, Steven McDonagh, Sotirios A.
Tsaftaris, Timothy Hospedales | CheXGenBench: A Unified Benchmark For Fidelity, Privacy and Utility of
Synthetic Chest Radiographs | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | We introduce CheXGenBench, a rigorous and multifaceted evaluation framework
for synthetic chest radiograph generation that simultaneously assesses
fidelity, privacy risks, and clinical utility across state-of-the-art
text-to-image generative models. Despite rapid advancements in generative AI
for real-world imagery, medical domain evaluations have been hindered by
methodological inconsistencies, outdated architectural comparisons, and
disconnected assessment criteria that rarely address the practical clinical
value of synthetic samples. CheXGenBench overcomes these limitations through
standardised data partitioning and a unified evaluation protocol comprising
over 20 quantitative metrics that systematically analyse generation quality,
potential privacy vulnerabilities, and downstream clinical applicability across
11 leading text-to-image architectures. Our results reveal critical
inefficiencies in the existing evaluation protocols, particularly in assessing
generative fidelity, leading to inconsistent and uninformative comparisons. Our
framework establishes a standardised benchmark for the medical AI community,
enabling objective and reproducible comparisons while facilitating seamless
integration of both existing and future generative models. Additionally, we
release a high-quality, synthetic dataset, SynthCheX-75K, comprising 75K
radiographs generated by the top-performing model (Sana 0.6B) in our benchmark
to support further research in this critical domain. Through CheXGenBench, we
establish a new state-of-the-art and release our framework, models, and
SynthCheX-75K dataset at https://raman1121.github.io/CheXGenBench/
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 16:59:17 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Dutt",
"Raman",
""
],
[
"Sanchez",
"Pedro",
""
],
[
"Yao",
"Yongchen",
""
],
[
"McDonagh",
"Steven",
""
],
[
"Tsaftaris",
"Sotirios A.",
""
],
[
"Hospedales",
"Timothy",
""
]
] |
2505.10497 | Iurii Medvedev | Iurii Medvedev and Nuno Goncalves | MorphGuard: Morph Specific Margin Loss for Enhancing Robustness to Face
Morphing Attacks | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Face recognition has evolved significantly with the advancement of deep
learning techniques, enabling its widespread adoption in various applications
requiring secure authentication. However, this progress has also increased its
exposure to presentation attacks, including face morphing, which poses a
serious security threat by allowing one identity to impersonate another.
Therefore, modern face recognition systems must be robust against such attacks.
In this work, we propose a novel approach for training deep networks for face
recognition with enhanced robustness to face morphing attacks. Our method
modifies the classification task by introducing a dual-branch classification
strategy that effectively handles the ambiguity in the labeling of face morphs.
This adaptation allows the model to incorporate morph images into the training
process, improving its ability to distinguish them from bona fide samples.
Our strategy has been validated on public benchmarks, demonstrating its
effectiveness in enhancing robustness against face morphing attacks.
Furthermore, our approach is universally applicable and can be integrated into
existing face recognition training pipelines to improve classification-based
recognition methods.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:00:16 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Medvedev",
"Iurii",
""
],
[
"Goncalves",
"Nuno",
""
]
] |
2505.10498 | Sakshi Arya | Sakshi Arya | Batched Nonparametric Bandits via k-Nearest Neighbor UCB | 25 pages, 6 figures | null | null | null | stat.ML cs.LG math.ST stat.ME stat.TH | http://creativecommons.org/licenses/by/4.0/ | We study sequential decision-making in batched nonparametric contextual
bandits, where actions are selected over a finite horizon divided into a small
number of batches. Motivated by constraints in domains such as medicine and
marketing -- where online feedback is limited -- we propose a nonparametric
algorithm that combines adaptive k-nearest neighbor (k-NN) regression with the
upper confidence bound (UCB) principle. Our method, BaNk-UCB, is fully
nonparametric, adapts to the context dimension, and is simple to implement.
Unlike prior work relying on parametric or binning-based estimators, BaNk-UCB
uses local geometry to estimate rewards and adaptively balances exploration and
exploitation. We provide near-optimal regret guarantees under standard
Lipschitz smoothness and margin assumptions, using a theoretically motivated
batch schedule that balances regret across batches and achieves minimax-optimal
rates. Empirical evaluations on synthetic and real-world datasets demonstrate
that BaNk-UCB consistently outperforms binning-based baselines.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:00:51 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Arya",
"Sakshi",
""
]
] |
2505.10499 | Mahadevan Subramanian | Mahadevan Subramanian, Guo Zheng, Liang Jiang | Achievable rates for concatenated square Gottesman-Kitaev-Preskill codes | 14+15 pages, 6+4 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | The Gottesman-Kitaev-Preskill (GKP) codes are known to achieve optimal rates
under displacement noise and pure loss channels, which establishes theoretical
foundations for its optimality. However, such optimal rates are only known to
be achieved at a discrete set of noise strength with the current self-dual
symplectic lattice construction. In this work, we develop a new coding strategy
using concatenated continuous variable - discrete variable encodings to go
beyond past results and establish GKP's optimal rate over all noise strengths.
In particular, for displacement noise, the rate is obtained through a
constructive approach by concatenating GKP codes with a quantum polar code and
analog decoding. For pure loss channel, we prove the existence of
capacity-achieving GKP codes through a random coding approach. These results
highlight the capability of concatenation-based GKP codes and provides new
methods for constructing good GKP lattices.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:01:51 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Subramanian",
"Mahadevan",
""
],
[
"Zheng",
"Guo",
""
],
[
"Jiang",
"Liang",
""
]
] |
2505.10500 | Adrien Lesage | Tu Duyen Nguyen, Adrien Lesage, Clotilde Cantini, Rachid Riad | Quantized Approximate Signal Processing (QASP): Towards Homomorphic
Encryption for audio | 34 pages, 5 figures | null | null | null | eess.AS cs.CR cs.SD | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Audio and speech data are increasingly used in machine learning applications
such as speech recognition, speaker identification, and mental health
monitoring. However, the passive collection of this data by audio listening
devices raises significant privacy concerns. Fully homomorphic encryption (FHE)
offers a promising solution by enabling computations on encrypted data and
preserving user privacy. Despite its potential, prior attempts to apply FHE to
audio processing have faced challenges, particularly in securely computing time
frequency representations, a critical step in many audio tasks.
Here, we addressed this gap by introducing a fully secure pipeline that
computes, with FHE and quantized neural network operations, four fundamental
time-frequency representations: Short-Time Fourier Transform (STFT), Mel
filterbanks, Mel-frequency cepstral coefficients (MFCCs), and gammatone
filters. Our methods also support the private computation of audio descriptors
and convolutional neural network (CNN) classifiers. Besides, we proposed
approximate STFT algorithms that lighten computation and bit use for
statistical and machine learning analyses.
We ran experiments on the VocalSet and OxVoc datasets demonstrating the fully
private computation of our approach. We showed significant performance
improvements with STFT approximation in private statistical analysis of audio
markers, and for vocal exercise classification with CNNs. Our results reveal
that our approximations substantially reduce error rates compared to
conventional STFT implementations in FHE. We also demonstrated a fully private
classification based on the raw audio for gender and vocal exercise
classification. Finally, we provided a practical heuristic for parameter
selection, making quantized approximate signal processing accessible to
researchers and practitioners aiming to protect sensitive audio data.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:01:52 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Nguyen",
"Tu Duyen",
""
],
[
"Lesage",
"Adrien",
""
],
[
"Cantini",
"Clotilde",
""
],
[
"Riad",
"Rachid",
""
]
] |
2505.10501 | Joshua N. Cooper | Joshua Cooper, Zhibin Du | Determinants of Steiner Distance Hypermatrices | 10 pages, 1 figure | null | null | null | math.CO | http://creativecommons.org/licenses/by/4.0/ | Generalizing work from the 1970s on the determinants of distance
hypermatrices of trees, we consider the hyperdeterminants of order-$k$ Steiner
distance hypermatrices of trees on $n$ vertices. We show that they can be
nearly diagonalized as $k$-forms, generalizing a result of Graham-Lov\'{a}sz,
implying a tensor version of ``conditional negative definiteness'', providing
new proofs of previous results of the authors and Tauscheck, and resolving the
conjecture that these hyperdeterminants depend only on $k$ and $n$ -- as
Graham-Pollak showed for $k=2$. We conclude with some open questions.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:04:21 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Cooper",
"Joshua",
""
],
[
"Du",
"Zhibin",
""
]
] |
2505.10502 | Yifan Gao | Yifan Gao, Yaoxian Dong, Wenbin Wu, Chaoyang Ge, Feng Yuan, Jiaxi
Sheng, Haoyue Li and Xin Gao | WeGA: Weakly-Supervised Global-Local Affinity Learning Framework for
Lymph Node Metastasis Prediction in Rectal Cancer | null | null | null | null | eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate lymph node metastasis (LNM) assessment in rectal cancer is essential
for treatment planning, yet current MRI-based evaluation shows unsatisfactory
accuracy, leading to suboptimal clinical decisions. Developing automated
systems also faces significant obstacles, primarily the lack of node-level
annotations. Previous methods treat lymph nodes as isolated entities rather
than as an interconnected system, overlooking valuable spatial and contextual
information. To solve this problem, we present WeGA, a novel weakly-supervised
global-local affinity learning framework that addresses these challenges
through three key innovations: 1) a dual-branch architecture with DINOv2
backbone for global context and residual encoder for local node details; 2) a
global-local affinity extractor that aligns features across scales through
cross-attention fusion; and 3) a regional affinity loss that enforces
structural coherence between classification maps and anatomical regions.
Experiments across one internal and two external test centers demonstrate that
WeGA outperforms existing methods, achieving AUCs of 0.750, 0.822, and 0.802
respectively. By effectively modeling the relationships between individual
lymph nodes and their collective context, WeGA provides a more accurate and
generalizable approach for lymph node metastasis prediction, potentially
enhancing diagnostic precision and treatment selection for rectal cancer
patients.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:05:21 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Gao",
"Yifan",
""
],
[
"Dong",
"Yaoxian",
""
],
[
"Wu",
"Wenbin",
""
],
[
"Ge",
"Chaoyang",
""
],
[
"Yuan",
"Feng",
""
],
[
"Sheng",
"Jiaxi",
""
],
[
"Li",
"Haoyue",
""
],
[
"Gao",
"Xin",
""
]
] |
2505.10503 | Sho Katayama | Sho Katayama and Yasuhito Miyamoto | Infinite multiplicity of positive solutions of an inhomogeneous
supercritical elliptic equation on $\mathbb{R}^N$ | 33 pages | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We are concerned with positive radial solutions of the inhomogeneous elliptic
equation $\Delta u+K(|x|)u^p+\mu f(|x|)=0$ on $\mathbb{R}^N$, where $N\ge 3$,
$\mu>0$ and $K$ and $f$ are nonnegative nontrivial functions. If $K(r)\sim
r^{\alpha}$, $\alpha>-2$, near $r=0$, $K(r)\sim r^{\beta}$, $\beta>-2$, near
$r=\infty$ and certain assumptions on $f$ are imposed, then the problem has a
unique positive radial singular solution for a certain range of $\mu$. We show
that existence of a positive radial singular solution is equivalent to
existence of infinitely many positive bounded solutions which are not uniformly
bounded, if $p$ is between the critical Sobolev exponent $p_S(\alpha)$ and
Joseph-Lundgren exponent $p_{JL}(\alpha)$. Using these theorems, we establish
existence of infinitely many positive bounded solutions which are not uniformly
bounded, for $p_S(\alpha)<p<p_{JL}(\alpha)$ if $K(r)=r^{-\alpha}$, $\alpha>-2$.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:07:10 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Katayama",
"Sho",
""
],
[
"Miyamoto",
"Yasuhito",
""
]
] |
2505.10504 | Juliette Bruce | Eran Assaf, Madeline Brandt, Juliette Bruce, Melody Chan, Raluca Vlad | Tropicalizations of locally symmetric varieties | 79 pages, 6 tables, 4 figures | null | null | null | math.AG math.AT math.CO math.KT math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper provides a rigorous study of tropicalizations of locally symmetric
varieties. We give applications beyond tropical geometry, to the cohomology of
moduli spaces as well as to the cohomology of arithmetic groups. We study two
cases in detail: the case of level structures on the moduli space
$\mathcal{A}_{g}$ of abelian varieties, and the special unitary case.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:09:27 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Assaf",
"Eran",
""
],
[
"Brandt",
"Madeline",
""
],
[
"Bruce",
"Juliette",
""
],
[
"Chan",
"Melody",
""
],
[
"Vlad",
"Raluca",
""
]
] |
2505.10505 | Aniruddha Seal | Aniruddha Seal, Simone Perego, Matthew R. Hennefarth, Umberto Raucci,
Luigi Bonati, Andrew L. Ferguson, Michele Parrinello, Laura Gagliardi | Weighted Active Space Protocol for Multireference Machine-Learned
Potentials | null | null | null | null | physics.chem-ph cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multireference methods such as multiconfiguration pair-density functional
theory (MC-PDFT) offer an effective means of capturing electronic correlation
in systems with significant multiconfigurational character. However, their
application to train machine learning-based interatomic potentials (MLPs) for
catalytic dynamics has been challenging due to the sensitivity of
multireference calculations to the underlying active space, which complicates
achieving consistent energies and gradients across diverse nuclear
configurations. To overcome this limitation, we introduce the Weighted
Active-Space Protocol (WASP), a systematic approach to assign a consistent
active space for a given system across uncorrelated configurations. By
integrating WASP with MLPs and enhanced sampling techniques, we propose a
data-efficient active learning cycle that enables the training of an MLP on
multireference data. We demonstrate the method on the TiC+-catalyzed C-H
activation of methane, a reaction that poses challenges for Kohn-Sham density
functional theory due to its significant multireference character. This
framework enables accurate and efficient modeling of catalytic dynamics,
establishing a new paradigm for simulating complex reactive processes beyond
the limits of conventional electronic-structure methods.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:09:43 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Seal",
"Aniruddha",
""
],
[
"Perego",
"Simone",
""
],
[
"Hennefarth",
"Matthew R.",
""
],
[
"Raucci",
"Umberto",
""
],
[
"Bonati",
"Luigi",
""
],
[
"Ferguson",
"Andrew L.",
""
],
[
"Parrinello",
"Michele",
""
],
[
"Gagliardi",
"Laura",
""
]
] |
2505.10506 | Zhaoyu Liu | Xuetao Ma, Zhaoyu Liu, Jiaqi Cai, Kenji Watanabe, Takashi Taniguchi,
Xiaodong Xu, Jiun-Haw Chu, Matthew Yankowitz | Giant elastoresistance in magic-angle twisted bilayer graphene | 21 pages, 16 figures | null | null | null | cond-mat.mes-hall cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Strongly correlated and topological phases in moir\'e materials are
exquisitely sensitive to lattice geometry at both atomic and superlattice
length scales. Twist angle, pressure, and strain directly modify the lattice,
and thus act as highly effective tuning parameters. Here we examine electrical
transport in twisted bilayer graphene subjected to continuous uniaxial strain.
Near the magic angle ($\approx 1.1^{\circ}$), devices exhibit a pronounced
elastoresistance that depends on band filling and temperature, with a gauge
factor more than two orders of magnitude larger than that of conventional
metals. In selected doping regimes the elastoresistance exhibits a
Curie-Weiss-like temperature divergence. We discuss possible microscopic
origins, including nematic fluctuations and enhanced electronic entropy from
fluctuating isospin moments. Our work establishes uniaxial strain as a
versatile probe of correlated physics in a moir\'e material.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:10:34 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Ma",
"Xuetao",
""
],
[
"Liu",
"Zhaoyu",
""
],
[
"Cai",
"Jiaqi",
""
],
[
"Watanabe",
"Kenji",
""
],
[
"Taniguchi",
"Takashi",
""
],
[
"Xu",
"Xiaodong",
""
],
[
"Chu",
"Jiun-Haw",
""
],
[
"Yankowitz",
"Matthew",
""
]
] |
2505.10507 | Benedikt Ebing | Benedikt Ebing, Goran Glava\v{s} | The Devil Is in the Word Alignment Details: On Translation-Based
Cross-Lingual Transfer for Token Classification Tasks | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Translation-based strategies for cross-lingual transfer XLT such as
translate-train -- training on noisy target language data translated from the
source language -- and translate-test -- evaluating on noisy source language
data translated from the target language -- are competitive XLT baselines. In
XLT for token classification tasks, however, these strategies include label
projection, the challenging step of mapping the labels from each token in the
original sentence to its counterpart(s) in the translation. Although word
aligners (WAs) are commonly used for label projection, the low-level design
decisions for applying them to translation-based XLT have not been
systematically investigated. Moreover, recent marker-based methods, which
project labeled spans by inserting tags around them before (or after)
translation, claim to outperform WAs in label projection for XLT. In this work,
we revisit WAs for label projection, systematically investigating the effects
of low-level design decisions on token-level XLT: (i) the algorithm for
projecting labels between (multi-)token spans, (ii) filtering strategies to
reduce the number of noisily mapped labels, and (iii) the pre-tokenization of
the translated sentences. We find that all of these substantially impact
translation-based XLT performance and show that, with optimized choices, XLT
with WA offers performance at least comparable to that of marker-based methods.
We then introduce a new projection strategy that ensembles translate-train and
translate-test predictions and demonstrate that it substantially outperforms
the marker-based projection. Crucially, we show that our proposed ensembling
also reduces sensitivity to low-level WA design choices, resulting in more
robust XLT for token classification tasks.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:10:50 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Ebing",
"Benedikt",
""
],
[
"Glavaš",
"Goran",
""
]
] |
2505.10508 | Sr{\dj}an Trifunovi\'c PhD | Sr{\dj}an Trifunovi\'c | Contact in fluid-plate interaction: formation and detachment | null | null | null | null | math.AP | http://creativecommons.org/licenses/by/4.0/ | In this paper, we study an interaction problem between an elastic plate and a
compressible viscous fluid located between the rigid bottom $z=0$ and the
plate. First, by utilizing the vertical fluid dissipation, we show that
$\ln\eta(t) \in L^1$ for any $t>0$ provided that $\ln\eta_0\in L^1$, ensuring
that additional plate contact can form only on a set of a measure zero. Then,
by utilizing the expanding capability of compressible fluid pressure, we show
that all contact has to detach in finite time provided that the source force
acting onto the plate is not pushing down excessively. Finally, we show that
contact at any point can be detached in any given time with a strong enough
source force localized around that point which is pulling the plate up. This is
the first result where detachment of contact is proven in fluid-structure
interaction.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:11:38 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Trifunović",
"Srđan",
""
]
] |
2505.10509 | Velmurugan S | Velmurugan S | Vanishing pairs of conjugacy classes for the symmetric group | Comments are welcome! | null | null | null | math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this short note, we classify pairs of conjugacy classes of the symmetric
group such that any non-linear irreducible character of the symmetric group
vanishes on at least one of them.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:13:23 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"S",
"Velmurugan",
""
]
] |
2505.10510 | Svenja Jedhoff | Svenja Jedhoff, Hadi Kutabi, Anne Meyer, Paul-Christian B\"urkner | Efficient Uncertainty Propagation in Bayesian Two-Step Procedures | null | null | null | null | stat.ME | http://creativecommons.org/licenses/by/4.0/ | Bayesian inference provides a principled framework for probabilistic
reasoning. If inference is performed in two steps, uncertainty propagation
plays a crucial role in accounting for all sources of uncertainty and
variability. This becomes particularly important when both aleatoric
uncertainty, caused by data variability, and epistemic uncertainty, arising
from incomplete knowledge or missing data, are present. Examples include
surrogate models and missing data problems. In surrogate modeling, the
surrogate is used as a simplified approximation of a resource-heavy and costly
simulation. The uncertainty from the surrogate-fitting process can be
propagated using a two-step procedure. For modeling with missing data, methods
like Multivariate Imputation by Chained Equations (MICE) generate multiple
datasets to account for imputation uncertainty. These approaches, however, are
computationally expensive, as multiple models must be fitted separately to
surrogate parameters respectively imputed datasets.
To address these challenges, we propose an efficient two-step approach that
reduces computational overhead while maintaining accuracy. By selecting a
representative subset of draws or imputations, we construct a mixture
distribution to approximate the desired posteriors using Pareto smoothed
importance sampling. For more complex scenarios, this is further refined with
importance weighted moment matching and an iterative procedure that broadens
the mixture distribution to better capture diverse posterior distributions.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:14:21 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Jedhoff",
"Svenja",
""
],
[
"Kutabi",
"Hadi",
""
],
[
"Meyer",
"Anne",
""
],
[
"Bürkner",
"Paul-Christian",
""
]
] |
2505.10511 | Victor Zheleznov | Victor Zheleznov, Stefan Bilbao, Alec Wright, Simon King | Learning Nonlinear Dynamics in Physical Modelling Synthesis using Neural
Ordinary Differential Equations | Accepted for publication in Proceedings of the 28th International
Conference on Digital Audio Effects (DAFx25), Ancona, Italy, September 2025 | null | null | null | cs.SD cs.LG eess.AS physics.comp-ph | http://creativecommons.org/licenses/by/4.0/ | Modal synthesis methods are a long-standing approach for modelling
distributed musical systems. In some cases extensions are possible in order to
handle geometric nonlinearities. One such case is the high-amplitude vibration
of a string, where geometric nonlinear effects lead to perceptually important
effects including pitch glides and a dependence of brightness on striking
amplitude. A modal decomposition leads to a coupled nonlinear system of
ordinary differential equations. Recent work in applied machine learning
approaches (in particular neural ordinary differential equations) has been used
to model lumped dynamic systems such as electronic circuits automatically from
data. In this work, we examine how modal decomposition can be combined with
neural ordinary differential equations for modelling distributed musical
systems. The proposed model leverages the analytical solution for linear
vibration of system's modes and employs a neural network to account for
nonlinear dynamic behaviour. Physical parameters of a system remain easily
accessible after the training without the need for a parameter encoder in the
network architecture. As an initial proof of concept, we generate synthetic
data for a nonlinear transverse string and show that the model can be trained
to reproduce the nonlinear dynamics of the system. Sound examples are
presented.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:17:21 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Zheleznov",
"Victor",
""
],
[
"Bilbao",
"Stefan",
""
],
[
"Wright",
"Alec",
""
],
[
"King",
"Simon",
""
]
] |
2505.10512 | Sara Najem | Sara Najem, Leonid Klushin, and Jihad Touma | Networked Infectiousness: Cascades, Power Laws, and Kinetics | null | null | null | null | cond-mat.stat-mech nlin.AO | http://creativecommons.org/licenses/by/4.0/ | Networked SIR models have become essential workhorses in the modeling of
epidemics, their inception, propagation and control. Here, and building on this
venerable tradition, we report on the emergence of a remarkable
self-organization of infectiousness in the wake of a propagating disease front.
It manifests as a cascading power-law distribution of disease strength in
networked SIR simulations, and is then confirmed with suitably defined
kinetics, then stochastic modeling of surveillance data. Given the success of
the networked SIR models which brought it to light, we expect this
scale-invariant feature to be of universal significance, characterizing the
evolution of disease within and across transportation networks, informing the
design of control strategies, and providing a litmus test for the soundness of
disease propagation models.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:19:18 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Najem",
"Sara",
""
],
[
"Klushin",
"Leonid",
""
],
[
"Touma",
"Jihad",
""
]
] |
2505.10513 | Surabhi Luthra | Surabhi Luthra, Alexandra E. Moylett, Dan E. Browne and Earl T.
Campbell | Unlocking early fault-tolerant quantum computing with mitigated magic
dilution | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | As quantum computing progresses towards the early fault-tolerant regime,
quantum error correction will play a crucial role in protecting qubits and
enabling logical Clifford operations. However, the number of logical qubits
will initially remain limited, posing challenges for resource-intensive tasks
like magic state distillation. It is therefore essential to develop efficient
methods for implementing non-Clifford operations, such as small-angle
rotations, to maximise the computational capabilities of devices within these
constraints. In this work, we introduce mitigated magic dilution (MMD) as an
approach to synthesise small-angle rotations by employing quantum error
mitigation techniques to sample logical Clifford circuits given noisy encoded
magic states. We explore the utility of our approach for the simulation of the
2D Fermi-Hubbard model. We identify evolution time regimes where MMD
outperforms the Ross-Selinger gate synthesis method [Quantum Inf.\ Comp.\
\textbf{16}, 901-953 (2016), arXiv:1403.2975] in the number of noisy encoded
magic states required for square lattices up to size $8 \times 8$. Moreover, we
demonstrate that our method can provide a practical advantage which is
quantified by a better-than-quadratic improvement in the resource requirements
for small-angle rotations over classical simulators. This work paves the way
for early fault-tolerant demonstrations on devices supporting millions of
quantum operations, the so-called MegaQuOp regime.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:19:19 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Luthra",
"Surabhi",
""
],
[
"Moylett",
"Alexandra E.",
""
],
[
"Browne",
"Dan E.",
""
],
[
"Campbell",
"Earl T.",
""
]
] |
2505.10514 | Jieqi Di | Jieqi Di and Sigr\'un Andrad\'ottir and Hayriye Ayhan | Optimal Pricing With Impatient Customers | null | null | null | null | math.OC | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We investigate the optimal pricing strategy in a service-providing framework,
where customers can become impatient and leave the system prior to service
completion. In this setting, a price is quoted to an incoming customer based on
the current number of customers in the system. When the quoted price is lower
than the price the incoming customer is willing to pay (which follows a fixed
probability distribution), then the customer joins the system and a reward
equal to the quoted price is earned. A cost is incurred upon abandonment and a
holding cost is incurred for customers waiting to be served. Our goal is to
determine the pricing policy that maximizes the long-run average profit. Unlike
traditional queueing systems without abandonments, we show that the optimal
quoted prices do not always increase with the queue length in this setting. In
particular, we prove that the optimal pricing policy is always uni-modal and
provide conditions guaranteeing that the optimal policy is increasing in the
number of customers in the system. Moreover, we introduce two heuristics that
simplify the optimal dynamic pricing policy. Both heuristics admit customers
until the number of customers in the system reaches a certain threshold. The
cutoff static policy charges all admitted customers a fixed price while the two
price policy charges one price when the arriving customer can enter service
immediately and another price if the customer needs to wait. By selecting the
price(s) and threshold that maximize the long-run average profit, both
heuristics achieve near optimality and the two price policy provides more
robustness compared to the cutoff static policy.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:20:28 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Di",
"Jieqi",
""
],
[
"Andradóttir",
"Sigrún",
""
],
[
"Ayhan",
"Hayriye",
""
]
] |
2505.10515 | Seongun Kim | Seongun Kim, Sol A Kim, Geonhyeong Kim, Enver Menadjiev, Chanwoo Lee,
Seongwook Chung, Nari Kim, Jaesik Choi | PnPXAI: A Universal XAI Framework Providing Automatic Explanations
Across Diverse Modalities and Models | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Recently, post hoc explanation methods have emerged to enhance model
transparency by attributing model outputs to input features. However, these
methods face challenges due to their specificity to certain neural network
architectures and data modalities. Existing explainable artificial intelligence
(XAI) frameworks have attempted to address these challenges but suffer from
several limitations. These include limited flexibility to diverse model
architectures and data modalities due to hard-coded implementations, a
restricted number of supported XAI methods because of the requirements for
layer-specific operations of attribution methods, and sub-optimal
recommendations of explanations due to the lack of evaluation and optimization
phases. Consequently, these limitations impede the adoption of XAI technology
in real-world applications, making it difficult for practitioners to select the
optimal explanation method for their domain. To address these limitations, we
introduce \textbf{PnPXAI}, a universal XAI framework that supports diverse data
modalities and neural network models in a Plug-and-Play (PnP) manner. PnPXAI
automatically detects model architectures, recommends applicable explanation
methods, and optimizes hyperparameters for optimal explanations. We validate
the framework's effectiveness through user surveys and showcase its versatility
across various domains, including medicine and finance.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:21:54 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Kim",
"Seongun",
""
],
[
"Kim",
"Sol A",
""
],
[
"Kim",
"Geonhyeong",
""
],
[
"Menadjiev",
"Enver",
""
],
[
"Lee",
"Chanwoo",
""
],
[
"Chung",
"Seongwook",
""
],
[
"Kim",
"Nari",
""
],
[
"Choi",
"Jaesik",
""
]
] |
2505.10516 | Vitaly Kresin | Marie Riddle, Angel I. Pena Dominguez, Benjamin S. Kamerin, Vitaly V.
Kresin | Shift of nanodroplet and nanocluster size distributions induced by
dopant pick-up statistics | 11 pages | Journal of Chemical Physics v.162, 054307 (2025) | 10.1063/5.0252959 10.1063/5.0252959 | null | physics.atm-clus physics.chem-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In pick-up experiments using nanodroplet and nanocluster beams, the size
distribution of hosts carrying a specified number of dopants changes when the
vapor density in the pick-up region is altered. This change, analyzed here, has
quantitative consequences for the interpretation of data that are sensitive to
host size, such as mass spectrometric, spectroscopic, and deflection
measurements.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:22:27 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Riddle",
"Marie",
""
],
[
"Dominguez",
"Angel I. Pena",
""
],
[
"Kamerin",
"Benjamin S.",
""
],
[
"Kresin",
"Vitaly V.",
""
]
] |
2505.10517 | Gerardo Chowell | Yuganthi R. Liyanage, Omar Saucedo, Necibe Tuncer, Gerardo Chowell | A Tutorial on Structural Identifiability of Epidemic Models Using
StructuralIdentifiability.jl | null | null | null | null | q-bio.QM | http://creativecommons.org/licenses/by/4.0/ | Structural identifiability -- the theoretical ability to uniquely recover
model parameters from ideal, noise-free data -- is a prerequisite for reliable
parameter estimation in epidemic modeling. Despite its importance, structural
identifiability analysis remains underutilized in the infectious disease
modeling literature. In this tutorial, we present a practical and reproducible
workflow for conducting structural identifiability analysis of ordinary
differential equation models using the Julia package
\texttt{StructuralIdentifiability.jl}. We apply the tool to a range of epidemic
models, including SEIR variants with asymptomatic and pre-symptomatic
transmission, vector-borne systems, and models incorporating hospitalization
and disease-induced mortality. We compare results from
StructuralIdentifiability.jl with those obtained using DAISY, a widely used
differential algebra tool, and highlight cases where the Julia package succeeds
in analyzing models that DAISY cannot handle. Our findings underscore how
identifiability depends on model structure, the availability of initial
conditions, and the choice of observed states. All code and diagrams are
publicly available, making this tutorial a valuable reference for researchers
and educators working with dynamic disease models.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:22:47 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Liyanage",
"Yuganthi R.",
""
],
[
"Saucedo",
"Omar",
""
],
[
"Tuncer",
"Necibe",
""
],
[
"Chowell",
"Gerardo",
""
]
] |
2505.10518 | Anastasios Gerontopoulos | Anastasios Gerontopoulos, Spyros Gidaris, Nikos Komodakis | Multi-Token Prediction Needs Registers | null | null | null | null | cs.CL cs.AI cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Multi-token prediction has emerged as a promising objective for improving
language model pretraining, but its benefits have not consistently generalized
to other settings such as fine-tuning. In this paper, we propose MuToR, a
simple and effective approach to multi-token prediction that interleaves
learnable register tokens into the input sequence, each tasked with predicting
future targets. Compared to existing methods, MuToR offers several key
advantages: it introduces only a negligible number of additional parameters,
requires no architectural changes--ensuring compatibility with off-the-shelf
pretrained language models--and remains aligned with the next-token pretraining
objective, making it especially well-suited for supervised fine-tuning.
Moreover, it naturally supports scalable prediction horizons. We demonstrate
the effectiveness and versatility of MuToR across a range of use cases,
including supervised fine-tuning, parameter-efficient fine-tuning (PEFT), and
pretraining, on challenging generative tasks in both language and vision
domains. Our code will be available at: https://github.com/nasosger/MuToR.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:25:03 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Gerontopoulos",
"Anastasios",
""
],
[
"Gidaris",
"Spyros",
""
],
[
"Komodakis",
"Nikos",
""
]
] |
2505.10519 | P. M. Aronow | P. M. Aronow, Austin Jang, Molly Offer-Westort | On the Foundations of the Design-Based Approach | null | null | null | null | stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The design-based paradigm may be adopted in causal queries and survey
sampling when we assume Rubin's stable unit treatment value assumption (SUTVA)
or impose similar frameworks. While often taken for granted, such assumptions
entail strong claims about the data generating process. We develop an
alternative design-based approach: we first invoke a generalized,
non-parametric model that allows for unrestricted forms of interference, such
as spillover. We define a new set of inferential targets and discuss their
interpretation under SUTVA and a weaker assumption that we call the No
Unmodeled Revealable Variation Assumption (NURVA). We then reconstruct the
standard paradigm, reconsidering SUTVA at the end rather than assuming it at
the beginning. Despite its similarity to SUTVA, we demonstrate the practical
insufficiency of NURVA in identifying substantively interesting quantities. In
so doing, we provide clarity on the nature and importance of SUTVA for applied
research.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:26:43 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Aronow",
"P. M.",
""
],
[
"Jang",
"Austin",
""
],
[
"Offer-Westort",
"Molly",
""
]
] |
2505.10520 | Ovidiu Munteanu | Ovidiu Munteanu and Jiaping Wang | Sharp integral bound of scalar curvature on $3$-manifolds | 21 pages | null | null | null | math.DG | http://creativecommons.org/licenses/by/4.0/ | It is shown that the integral of the scalar curvature on a geodesic ball of
radius $R$ in a three-dimensional complete manifold with nonnegative Ricci
curvature is bounded above by $8\pi R$ asymptotically for large $R$ provided
that the scalar curvature is bounded between two positive constants.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:29:18 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Munteanu",
"Ovidiu",
""
],
[
"Wang",
"Jiaping",
""
]
] |
2505.10521 | Pavel Shumyatsky | Eloisa Detomi, Robert M. Guralnick, Marta Morigi, Pavel Shumyatsky | Commuting probability for conjugate subgroups of a finite group | null | null | null | null | math.GR | http://creativecommons.org/licenses/by/4.0/ | Given two subgroups H,K of a finite group G, the probability that a pair of
random elements from H and K commutes is denoted by \pr(H,K). We address the
following question.
Let P be a p-subgroup of a finite group G and assume that \pr(P,P^x)\geq\e>0
for every x\in G. Is the order of P modulo O_p(G) bounded in terms of e only?
With respect to this question, we establish several positive results but show
that in general the answer is negative. In particular, we prove that if the
composition factors of G which are isomorphic to simple groups of Lie type in
characteristic p, have Lie rank at most n, then the order of P modulo O_p(G) is
bounded in terms of n and e only. If P is a Sylow p-subgroup of G, then the
order of P modulo O_p(G) is bounded in terms e only. Some other results of
similar flavour are established.
We also show that if \pr(P_1,P_2)>0 for every two Sylow p-subgroups P_1,P_2
of a profinite group G, then O_{p,p'}(G) is open in G.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:30:28 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Detomi",
"Eloisa",
""
],
[
"Guralnick",
"Robert M.",
""
],
[
"Morigi",
"Marta",
""
],
[
"Shumyatsky",
"Pavel",
""
]
] |
2505.10522 | Xinrui Wang | Xinrui Wang, Yan Jin | Knowledge capture, adaptation and composition (KCAC): A framework for
cross-task curriculum learning in robotic manipulation | null | null | null | null | cs.RO cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Reinforcement learning (RL) has demonstrated remarkable potential in robotic
manipulation but faces challenges in sample inefficiency and lack of
interpretability, limiting its applicability in real world scenarios. Enabling
the agent to gain a deeper understanding and adapt more efficiently to diverse
working scenarios is crucial, and strategic knowledge utilization is a key
factor in this process. This paper proposes a Knowledge Capture, Adaptation,
and Composition (KCAC) framework to systematically integrate knowledge transfer
into RL through cross-task curriculum learning. KCAC is evaluated using a two
block stacking task in the CausalWorld benchmark, a complex robotic
manipulation environment. To our knowledge, existing RL approaches fail to
solve this task effectively, reflecting deficiencies in knowledge capture. In
this work, we redesign the benchmark reward function by removing rigid
constraints and strict ordering, allowing the agent to maximize total rewards
concurrently and enabling flexible task completion. Furthermore, we define two
self-designed sub-tasks and implement a structured cross-task curriculum to
facilitate efficient learning. As a result, our KCAC approach achieves a 40
percent reduction in training time while improving task success rates by 10
percent compared to traditional RL methods. Through extensive evaluation, we
identify key curriculum design parameters subtask selection, transition timing,
and learning rate that optimize learning efficiency and provide conceptual
guidance for curriculum based RL frameworks. This work offers valuable insights
into curriculum design in RL and robotic learning.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:30:29 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Wang",
"Xinrui",
""
],
[
"Jin",
"Yan",
""
]
] |
2505.10523 | Vitaly Kresin | Thomas H. Villers, Benjamin S. Kamerin, Vitaly V. Kresin | Magnetic deflection of high-spin sodium dimers formed on helium
nanodroplets | Published as "Editors' Choice" | Journal of Physical Chemistry Letters 16, 4436 (2025) | 10.1021/acs.jpclett.5c00861 | null | physics.atm-clus physics.atom-ph physics.chem-ph quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Spectroscopic data on alkali-atom dimers residing on the surface of liquid
helium nanodroplets have revealed that they are detected primarily in the
weakly bound, metastable, spin-triplet state. Here, by measuring the magnetic
Stern-Gerlach deflection of a sodium-doped nanodroplet beam, we transparently
demonstrate the abundance of high-magnetic-moment dimers. Their electron spins
thermalize with the cryogenic superfluid droplets and become fully oriented by
the external magnetic field.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:31:10 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Villers",
"Thomas H.",
""
],
[
"Kamerin",
"Benjamin S.",
""
],
[
"Kresin",
"Vitaly V.",
""
]
] |
2505.10524 | MuXin Liu | Muxin Liu, Lile Wang and Peng Peng | AGN star dynamics under the Influence of Outflow-Ambient Interactions | 15 pages, 13 figures | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Stars with outflows interacting with ambient gas experience accelerations
arising from the gravitational feedback induced by the interaction structure.
In this work, three-dimensional (3D) local shearing box simulations are
performed to investigate the dynamical evolution of a star with outflows
embedded in the outer regions of an active galactic nucleus (AGN) disk. Two
types of stellar wind are considered: isotropic winds and axisymmetric jets,
along with variations in the radial pressure gradient profile. The results show
that anti-friction enables AGN stars to acquire angular momentum from the
ambient gas, resulting in outward migration away from the disk center. The
formation and stability of the head-wind structure, which is crucial for
maintaining anti-friction, are sensitive to both the strength of the stellar
outflow and the radial pressure gradient of the disk gas. Once the head-wind
structure is disrupted, the anti-friction effect ceases to operate effectively.
A case study is also presented, focusing on a stellar-mass black hole (sBH) in
an AGN disk. It is shown that jet material launched along the z-axis is
confined to the trailing side of the object's motion by high gas inflow
velocities, thereby activating anti-friction and inducing outward migration. If
such an sBH migrates inward initially, the interplay between inward and outward
migration may trap it at an equilibrium radius, potentially facilitating the
formation and merger of black hole binaries.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:34:30 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Liu",
"Muxin",
""
],
[
"Wang",
"Lile",
""
],
[
"Peng",
"Peng",
""
]
] |
2505.10525 | Jeremy Tyson | Jonathan M. Fraser and Jeremy T. Tyson | Sobolev and quasiconformal distortion of intermediate dimension with
applications to conformal dimension | 51 pages | null | null | null | math.MG math.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the distortion of intermediate dimension under supercritical Sobolev
mappings and also under quasiconformal or quasisymmetric homeomorphisms. In
particular, we extend to the setting of intermediate dimensions both the
Gehring--V\"ais\"al\"a theorem on dilatation-dependent quasiconformal
distortion of dimension and Kovalev's theorem on the nonexistence of metric
spaces with conformal dimension strictly between zero and one. Applications
include new contributions to the quasiconformal classification of Euclidean
sets and a new sufficient condition for the vanishing of conformal box-counting
dimension. We illustrate our conclusions with specific consequences for
Bedford--McMullen carpets, samples of Mandelbrot percolation, and product sets
containing a polynomially convergent sequence factor.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:36:06 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Fraser",
"Jonathan M.",
""
],
[
"Tyson",
"Jeremy T.",
""
]
] |
2505.10526 | Vithursan Thangarasa | Mugilan Ganesan, Shane Segal, Ankur Aggarwal, Nish Sinnadurai, Sean
Lie, Vithursan Thangarasa | MASSV: Multimodal Adaptation and Self-Data Distillation for Speculative
Decoding of Vision-Language Models | Main paper: 11 pp., 4 figs., 3 tabs.; Supplementary: 2 pp | null | null | null | cs.LG cs.CL cs.CV | http://creativecommons.org/licenses/by/4.0/ | Speculative decoding significantly accelerates language model inference by
enabling a lightweight draft model to propose multiple tokens that a larger
target model verifies simultaneously. However, applying this technique to
vision-language models (VLMs) presents two fundamental challenges: small
language models that could serve as efficient drafters lack the architectural
components to process visual inputs, and their token predictions fail to match
those of VLM target models that consider visual context. We introduce
Multimodal Adaptation and Self-Data Distillation for Speculative Decoding of
Vision-Language Models (MASSV), which transforms existing small language models
into effective multimodal drafters through a two-phase approach. MASSV first
connects the target VLM's vision encoder to the draft model via a lightweight
trainable projector, then applies self-distilled visual instruction tuning
using responses generated by the target VLM to align token predictions.
Comprehensive experiments across the Qwen2.5-VL and Gemma3 model families
demonstrate that MASSV increases accepted length by up to 30% and delivers
end-to-end inference speedups of up to 1.46x on visually-grounded tasks. MASSV
provides a scalable, architecture-compatible method for accelerating both
current and future VLMs.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:37:00 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Ganesan",
"Mugilan",
""
],
[
"Segal",
"Shane",
""
],
[
"Aggarwal",
"Ankur",
""
],
[
"Sinnadurai",
"Nish",
""
],
[
"Lie",
"Sean",
""
],
[
"Thangarasa",
"Vithursan",
""
]
] |
2505.10527 | Binghai Wang | Binghai Wang, Runji Lin, Keming Lu, Le Yu, Zhenru Zhang, Fei Huang,
Chujie Zheng, Kai Dang, Yang Fan, Xingzhang Ren, An Yang, Binyuan Hui,
Dayiheng Liu, Tao Gui, Qi Zhang, Xuanjing Huang, Yu-Gang Jiang, Bowen Yu,
Jingren Zhou, Junyang Lin | WorldPM: Scaling Human Preference Modeling | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Motivated by scaling laws in language modeling that demonstrate how test loss
scales as a power law with model and dataset sizes, we find that similar laws
exist in preference modeling. We propose World Preference Modeling$ (WorldPM)
to emphasize this scaling potential, where World Preference embodies a unified
representation of human preferences. In this paper, we collect preference data
from public forums covering diverse user communities, and conduct extensive
training using 15M-scale data across models ranging from 1.5B to 72B
parameters. We observe distinct patterns across different evaluation metrics:
(1) Adversarial metrics (ability to identify deceptive features) consistently
scale up with increased training data and base model size; (2) Objective
metrics (objective knowledge with well-defined answers) show emergent behavior
in larger language models, highlighting WorldPM's scalability potential; (3)
Subjective metrics (subjective preferences from a limited number of humans or
AI) do not demonstrate scaling trends. Further experiments validate the
effectiveness of WorldPM as a foundation for preference fine-tuning. Through
evaluations on 7 benchmarks with 20 subtasks, we find that WorldPM broadly
improves the generalization performance across human preference datasets of
varying sizes (7K, 100K and 800K samples), with performance gains exceeding 5%
on many key subtasks. Integrating WorldPM into our internal RLHF pipeline, we
observe significant improvements on both in-house and public evaluation sets,
with notable gains of 4% to 8% in our in-house evaluations.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:38:37 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Wang",
"Binghai",
""
],
[
"Lin",
"Runji",
""
],
[
"Lu",
"Keming",
""
],
[
"Yu",
"Le",
""
],
[
"Zhang",
"Zhenru",
""
],
[
"Huang",
"Fei",
""
],
[
"Zheng",
"Chujie",
""
],
[
"Dang",
"Kai",
""
],
[
"Fan",
"Yang",
""
],
[
"Ren",
"Xingzhang",
""
],
[
"Yang",
"An",
""
],
[
"Hui",
"Binyuan",
""
],
[
"Liu",
"Dayiheng",
""
],
[
"Gui",
"Tao",
""
],
[
"Zhang",
"Qi",
""
],
[
"Huang",
"Xuanjing",
""
],
[
"Jiang",
"Yu-Gang",
""
],
[
"Yu",
"Bowen",
""
],
[
"Zhou",
"Jingren",
""
],
[
"Lin",
"Junyang",
""
]
] |
2505.10528 | Lingxin Meng | L. Meng, R. Bates, C. Buttar, G. Calderini, F. Crescioli, L.
Cunningham, Y. Dieter, R. Han, T. Heim, S. Hirose, F. Huegging, C. Hultquist,
D. Kim, A. Korn, M. Marjanovic, J. Metcalfe, K. Nakamura, J. Pater, H.
Pernegger, M.A.A. Samy, M. Schuessler, A. Sharma, E. Thompson, M. Backhaus,
J. Christiansen, F. Loddon | Pixel column issue in the ATLAS Inner Tracker modules | 8 pages, 11 figures, 20th Anniversary Trento Workshop on Advanced
Silicon Radiation Detectors 4--6 February 2025 | null | null | ATL-COM-ITK-2025-043 | physics.ins-det hep-ex | http://creativecommons.org/licenses/by/4.0/ | Pixel modules are currently being built for the ATLAS ITk Pixel detector
upgrade. During the preproduction phase, recurring chip malfunctioning was
observed during electrical testing. It was possible to bypass this issue by
disabling some pixel core columns in the ITkPix readout chip. Therefore the
issue is called "core column issue" which is a direct disqualifier for a pixel
module. A concerning number of cases has been observed in pixel modules with
ITkPix v1.1 as well as v2 chips which significantly impacts the module yield.
However, the behaviour is erratic and there is not any evidence hinting at the
origin of this issue. These proceedings outline the investigations of the
issue, highlighting the electrical behaviour during testing, present findings
from the data collected via our production database and through visual
inspection, and point towards possible causes of the issue.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:38:55 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Meng",
"L.",
""
],
[
"Bates",
"R.",
""
],
[
"Buttar",
"C.",
""
],
[
"Calderini",
"G.",
""
],
[
"Crescioli",
"F.",
""
],
[
"Cunningham",
"L.",
""
],
[
"Dieter",
"Y.",
""
],
[
"Han",
"R.",
""
],
[
"Heim",
"T.",
""
],
[
"Hirose",
"S.",
""
],
[
"Huegging",
"F.",
""
],
[
"Hultquist",
"C.",
""
],
[
"Kim",
"D.",
""
],
[
"Korn",
"A.",
""
],
[
"Marjanovic",
"M.",
""
],
[
"Metcalfe",
"J.",
""
],
[
"Nakamura",
"K.",
""
],
[
"Pater",
"J.",
""
],
[
"Pernegger",
"H.",
""
],
[
"Samy",
"M. A. A.",
""
],
[
"Schuessler",
"M.",
""
],
[
"Sharma",
"A.",
""
],
[
"Thompson",
"E.",
""
],
[
"Backhaus",
"M.",
""
],
[
"Christiansen",
"J.",
""
],
[
"Loddon",
"F.",
""
]
] |
2505.10529 | Alex Lukyanov | Hanan Hozan and Mike Baines and Alex Lukyanov | Oil slicks dynamics in semi-confined geometries in the presence of water
waves: analytical and numerical insights | 17 pages and 5 figures | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | We have shown previously that the dynamics of isolated oil slicks on the
water's surface after spills is significantly influenced by surface-wave motion
practically at the onset of the spreading process. In this work, we draw our
attention to another practical scenario of the oil slick's behaviour in
semi-confined geometries under the action of surface waves and in the presence
of oil leakage source. We hypothesize that the geometric constraints should
qualitatively change the oil layer response to the wave motion leading to
localization of oil spill domains even at modest levels of the surface-wave
perturbations. Several realistic cases were rigorously explored, with special
attention paid to the interplay, the combined effect of the two factors, oil
influx and wave motion. It was demonstrated quantitatively how the spreading
process can be either facilitated or suppressed leading to potential
confinement.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:40:04 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Hozan",
"Hanan",
""
],
[
"Baines",
"Mike",
""
],
[
"Lukyanov",
"Alex",
""
]
] |
2505.10530 | Benedikt Fauseweh PhD | Yanick S. Kind, Benedikt Fauseweh | Exploring Variational Entanglement Hamiltonians | 24 pages, 18 figures | null | null | null | quant-ph cond-mat.str-el | http://creativecommons.org/licenses/by/4.0/ | Recent advances in analog and digital quantum-simulation platforms have
enabled exploration of the spectrum of entanglement Hamiltonians via
variational algorithms. In this work we analyze the convergence properties of
the variationally obtained solutions and compare them to numerically exact
calculations in quantum critical systems. We demonstrate that interpreting the
cost functional as an integral permits the deployment of iterative quadrature
schemes, thereby reducing the required number of measurements by several orders
of magnitude. We further show that a modified ansatz captures deviations from
the Bisognano-Wichmann form in lattice models, improves convergence, and
provides a cost-function-level diagnostic for quantum phase transitions.
Finally, we establish that a low cost value does not by itself guarantee
convergence in trace distance. Nevertheless, it faithfully reproduces
degeneracies and spectral gaps, which are essential for applications to
topological phases.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:41:14 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Kind",
"Yanick S.",
""
],
[
"Fauseweh",
"Benedikt",
""
]
] |
2505.10531 | Hossein Hosseinabadi | Hossein Hosseinabadi, Yaroslav Tserkovnyak, Eugene Demler, Jamir
Marino | Magnon Nesting in Driven Two-Dimensional Quantum Magnets | 16 pages, 10 figures | null | null | null | cond-mat.mtrl-sci cond-mat.quant-gas nlin.PS quant-ph | http://creativecommons.org/licenses/by/4.0/ | We uncover a new class of dynamical quantum instability in driven magnets
leading to emergent enhancement of antiferromagnetic correlations even for
purely ferromagnetic microscopic couplings. A primary parametric amplification
creates a frequency-tuned nested magnon distribution in momentum space, which
seeds a secondary instability marked by the emergence of enhanced
antiferromagnetic correlations, mirroring Fermi surface nesting instabilities
in electronic systems. In sharp contrast to the fermionic case, however, the
magnon-driven instability is intrinsically non-equilibrium and fundamentally
inaccessible in thermal physics. Its quantum mechanical origin sets it apart
from classical instabilities such as Faraday and modulation instabilities,
which underlie several instances of dynamical behavior observed in magnetic and
cold-atom systems.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:41:44 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Hosseinabadi",
"Hossein",
""
],
[
"Tserkovnyak",
"Yaroslav",
""
],
[
"Demler",
"Eugene",
""
],
[
"Marino",
"Jamir",
""
]
] |
2505.10532 | Marcello Ortaggio | David Koko\v{s}ka, Marcello Ortaggio | On the uniqueness of the Kerr-(A)dS metric as a type II(D) solution in
six dimensions | 24 pages | null | null | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the class of six-dimensional $\Lambda$-vacuum spacetimes which admit
a non-degenerate multiple Weyl aligned null direction l (thus being of Weyl
type~II or more special) with a ``generic'' optical matrix. Subject to an
additional assumption on the asymptotic fall-off of the Weyl tensor, we obtain
the most general metric of this class, which is specified by one discrete
(normalized) and three continuous parameters. All solutions turn out to be
Kerr--Schild spacetimes of type~D and, in passing, we comment on their
Kerr--Schild double copy. We further show that the obtained family is locally
isometric to the general doubly-spinning Kerr-NUT-(A)dS metric with the NUTs
parameters switched off. In particular, the Kerr-(A)dS subclass and its
extensions (i.e., analytic continuation and ``infinite-rotation'' limit) are
recovered when certain polynomial metric functions are assumed to be fully
factorized. As a side result, a unified metric form which encompasses all three
branches of the extended Kerr-(A)dS family in all even dimensions is presented
in an appendix.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:41:50 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Kokoška",
"David",
""
],
[
"Ortaggio",
"Marcello",
""
]
] |
2505.10533 | Aaryan Sharma | Aaryan Sharma, Shivansh Gupta, Samar Agarwal, Vishak Prasad C., Ganesh
Ramakrishnan | Enhancing Multi-Image Question Answering via Submodular Subset Selection | null | null | null | null | cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Large multimodal models (LMMs) have achieved high performance in
vision-language tasks involving single image but they struggle when presented
with a collection of multiple images (Multiple Image Question Answering
scenario). These tasks, which involve reasoning over large number of images,
present issues in scalability (with increasing number of images) and retrieval
performance. In this work, we propose an enhancement for retriever framework
introduced in MIRAGE model using submodular subset selection techniques. Our
method leverages query-aware submodular functions, such as GraphCut, to
pre-select a subset of semantically relevant images before main retrieval
component. We demonstrate that using anchor-based queries and augmenting the
data improves submodular-retriever pipeline effectiveness, particularly in
large haystack sizes.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:41:52 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Sharma",
"Aaryan",
""
],
[
"Gupta",
"Shivansh",
""
],
[
"Agarwal",
"Samar",
""
],
[
"C.",
"Vishak Prasad",
""
],
[
"Ramakrishnan",
"Ganesh",
""
]
] |
2505.10534 | Suvashis Maity | Suvashis Maity | ACT-ing on inflation: Implications of non Bunch-Davies initial condition
and reheating on single-field slow roll models | 12 pages, 2 figures | null | null | null | astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate a class of slow roll inflationary models in the light of the
recent Cosmic Microwave Background constraints from Planck 2018, ACT DR6, DESI
DR1, and BICEP/\textit{Keck} 2018. The combined dataset favors a higher value
of the scalar spectral index $n_{_\mathrm{S}}=0.9743 \pm 0.0034$, which places
increased pressure on several conventional inflationary scenarios. In this
study, we analyze the observational viability of various well-motivated models,
including the $\alpha$-attractor E- and T-models, chaotic inflation, hilltop
inflation, and natural inflation. We incorporate the effects of a
post-inflationary phase of reheating and examine how the dynamics of reheating
influence the predictions in the $n_{_\mathrm{S}}-r$ plane. We also impose a
lower bound on the reheating temperature based on the constraint from the
effective number of relativistic species ($\Delta{N}_{\mathrm{eff}}$) arising
from primordial gravitational waves. While reheating improves agreement with
observations for some models, significant regions of parameter space remain
disfavored. Finally, we explore the impact of a non Bunch-Davies initial state
and demonstrate that it can substantially improve the fit to the
$n_{_\mathrm{S}}-r$ data across a broader class of inflationary models, thereby
offering a potentially viable mechanism for reconciling theory with the latest
observations.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:45:05 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Maity",
"Suvashis",
""
]
] |
2505.10535 | Matthew Green | Matthew J. Green, Jan van Roestel, and Tin Long Sunny Wong | A Catalogue of Ultracompact Mass-Transferring White Dwarf Binaries | Submitted to A&A on 03 March 2025, re-submitted to A&A on 15 May
following one round of reviewer comments. 14 pages of main text, 13 figures.
Comments welcome | null | null | null | astro-ph.SR astro-ph.GA astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present an overview and catalogue of all known mass-transferring binary
systems that have white dwarf accretors and orbital periods shorter than 70
minutes, including the AM CVn-type binary systems and a number of related
objects. In this paper, we provide an overview of the literature background and
describe recent breakthroughs, open questions, and connections to other fields.
This overview is intended to provide a starting point for researchers who are
relatively new to the fields of ultracompact binaries or AM CVn-type binaries.
We define the scope of the catalogue (the criteria for inclusion) and describe
its format and contents (what data have been included for each target). We also
summarise the observed properties of the known mass-transferring ultracompact
binaries, including their orbital periods and component stellar masses and
radii. The catalogue presented here aims to provide a central resource to
enable researchers to track the current status of the known sample. At the time
of writing, the catalogue includes 123 confirmed binaries and 29 candidates.
The catalogue is hosted on a public platform (Zenodo) with version control, and
will be updated regularly.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:46:42 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Green",
"Matthew J.",
""
],
[
"van Roestel",
"Jan",
""
],
[
"Wong",
"Tin Long Sunny",
""
]
] |
2505.10536 | Mohammad Ghalavand | Mohammad Ghalavand, Javad Hatami, Seyed Kamaledin Setarehdan and
Hananeh Ghalavand | Real-World fNIRS-Based Brain-Computer Interfaces: Benchmarking Deep
Learning and Classical Models in Interactive Gaming | null | null | null | null | q-bio.NC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Brain-Computer Interfaces enable direct communication between the brain and
external systems, with functional Near-Infrared Spectroscopy emerging as a
portable and non-invasive method for capturing cerebral hemodynamics. This
study investigates the classification of rest and task states during a
realistic, interactive tennis simulation using fNIRS signals and a range of
machine learning approaches. We benchmarked traditional classifiers based on
engineered features, Long Short-Term Memory networks on raw time-series data,
and Convolutional Neural Networks applied to Gramian Angular Field-transformed
images. Ensemble models like Extra Trees and Gradient Boosting achieved
accuracies above 97 percent, while the ResNet-based CNN reached 95.0 percent
accuracy with a near-perfect AUC of 99.2 percent, outperforming both LSTM and
EfficientNet architectures. A novel data augmentation strategy was employed to
equalize trial durations while preserving physiological integrity. Feature
importance analyses revealed that both oxygenated and deoxygenated hemoglobin
signals, particularly slope and RMS metrics, were key contributors to
classification performance. These findings demonstrate the strong potential of
fNIRS-based BCIs for deployment in dynamic, real-world environments and
underscore the advantages of deep learning models in decoding complex neural
signals.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:47:16 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Ghalavand",
"Mohammad",
""
],
[
"Hatami",
"Javad",
""
],
[
"Setarehdan",
"Seyed Kamaledin",
""
],
[
"Ghalavand",
"Hananeh",
""
]
] |
2505.10537 | Andrea Lacava | Filippo Olimpieri, Noemi Giustini, Andrea Lacava, Salvatore D'Oro,
Tommaso Melodia and Francesca Cuomo | LibIQ: Toward Real-Time Spectrum Classification in O-RAN dApps | 6 pages, 5 figures, 2 tables | null | null | null | cs.NI cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The O-RAN architecture is transforming cellular networks by adopting RAN
softwarization and disaggregation concepts to enable data-driven monitoring and
control of the network. Such management is enabled by RICs, which facilitate
near-real-time and non-real-time network control through xApps and rApps.
However, they face limitations, including latency overhead in data exchange
between the RAN and RIC, restricting real-time monitoring, and the inability to
access user plain data due to privacy and security constraints, hindering use
cases like beamforming and spectrum classification. In this paper, we leverage
the dApps concept to enable real-time RF spectrum classification with LibIQ, a
novel library for RF signals that facilitates efficient spectrum monitoring and
signal classification by providing functionalities to read I/Q samples as
time-series, create datasets and visualize time-series data through plots and
spectrograms. Thanks to LibIQ, I/Q samples can be efficiently processed to
detect external RF signals, which are subsequently classified using a CNN
inside the library. To achieve accurate spectrum analysis, we created an
extensive dataset of time-series-based I/Q samples, representing distinct
signal types captured using a custom dApp running on a 5G deployment over the
Colosseum network emulator and an OTA testbed. We evaluate our model by
deploying LibIQ in heterogeneous scenarios with varying center frequencies,
time windows, and external RF signals. In real-time analysis, the model
classifies the processed I/Q samples, achieving an average accuracy of
approximately 97.8\% in identifying signal types across all scenarios. We
pledge to release both LibIQ and the dataset created as a publicly available
framework upon acceptance.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:47:30 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Olimpieri",
"Filippo",
""
],
[
"Giustini",
"Noemi",
""
],
[
"Lacava",
"Andrea",
""
],
[
"D'Oro",
"Salvatore",
""
],
[
"Melodia",
"Tommaso",
""
],
[
"Cuomo",
"Francesca",
""
]
] |
2505.10538 | Laurie Williams | Imranur Rahman, Yasemin Acar, Michel Cukier, William Enck, Christian
Kastner, Alexandros Kapravelos, Dominik Wermke, Laurie Williams | S3C2 Summit 2024-09: Industry Secure Software Supply Chain Summit | null | null | null | null | cs.CR | http://creativecommons.org/licenses/by-sa/4.0/ | While providing economic and software development value, software supply
chains are only as strong as their weakest link. Over the past several years,
there has been an exponential increase in cyberattacks, specifically targeting
vulnerable links in critical software supply chains. These attacks disrupt the
day-to-day functioning and threaten the security of nearly everyone on the
internet, from billion-dollar companies and government agencies to hobbyist
open-source developers. The ever-evolving threat of software supply chain
attacks has garnered interest from the software industry and the US government
in improving software supply chain security.
On September 20, 2024, three researchers from the NSF-backed Secure Software
Supply Chain Center (S3C2) conducted a Secure Software Supply Chain Summit with
a diverse set of 12 practitioners from 9 companies. The goals of the Summit
were to: (1) to enable sharing between individuals from different companies
regarding practical experiences and challenges with software supply chain
security, (2) to help form new collaborations, (3) to share our observations
from our previous summits with industry, and (4) to learn about practitioners'
challenges to inform our future research direction. The summit consisted of
discussions of six topics relevant to the companies represented, including
updating vulnerable dependencies, component and container choice, malicious
commits, building infrastructure, large language models, and reducing entire
classes of vulnerabilities.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:48:14 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Rahman",
"Imranur",
""
],
[
"Acar",
"Yasemin",
""
],
[
"Cukier",
"Michel",
""
],
[
"Enck",
"William",
""
],
[
"Kastner",
"Christian",
""
],
[
"Kapravelos",
"Alexandros",
""
],
[
"Wermke",
"Dominik",
""
],
[
"Williams",
"Laurie",
""
]
] |
2505.10539 | Madhusudhan Nikku | Lorenzo Pica-Ciamarra, Nikku Madhusudhan, Gregory J. Cooke, Savvas
Constantinou and Martin Binet | A Systematic Search for Trace Molecules in Exoplanet K2-18 b | Submitted for publication | null | null | null | astro-ph.EP | http://creativecommons.org/licenses/by/4.0/ | The first transmission spectrum of the habitable-zone sub-Neptune K2-18 b
with JWST has opened a new avenue for atmospheric characterisation of temperate
low-mass exoplanets. The observations led to inferences of methane and carbon
dioxide, as well as of dimethyl sulfide (DMS) and/or dimethyl disulfide (DMDS),
both potential biosignatures. However, robust identification of DMS and/or DMDS
requires further observations to increase the detection significances. More
theoretical studies are also needed to identify potential false positives and
possible abiotic sources for these molecules. In the present work we
demonstrate the next step in this direction with a comprehensive and agnostic
search for other chemical species in the atmosphere of K2-18 b. Our exploration
includes 650 molecules, spanning a wide range of trace gases, including biotic,
abiotic, and anthropogenic gases on Earth. We investigate possible evidence for
any of these gases using three metrics: (a) evidence in the JWST mid-infrared
spectrum, (b) evidence in the JWST near-infrared spectrum, and (c) plausible
sources of production. We find three molecules, including DMS, which appear
promising across the datasets considered. The two molecules besides DMS are
diethyl sulfide and methyl acrylonitrile, which are more complex than DMS,
biogenic on Earth, and have no significant sources known beyond Earth. A few
other gases also provide comparable fits to a subset of the data considered but
again with limited known plausible sources. Our study highlights the need for
further observations to distinguish between possible trace gases in K2-18 b and
theoretical work to establish their plausible sources if confirmed on this
planet.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:48:51 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Pica-Ciamarra",
"Lorenzo",
""
],
[
"Madhusudhan",
"Nikku",
""
],
[
"Cooke",
"Gregory J.",
""
],
[
"Constantinou",
"Savvas",
""
],
[
"Binet",
"Martin",
""
]
] |
2505.10540 | Karl Blodgett | Karl N. Blodgett, Saumitra S. Phatak, Meng Raymond Chen, David Peana,
Claire Pritts, Jonathan D. Hood | Narrow-Line Electric Quadrupole Cooling And Background-Free Imaging Of A
Single Cs Atom With Spatially Structured Light | 19 pages, 5 main text figures, 8 appendix figures | null | null | null | physics.atom-ph quant-ph | http://creativecommons.org/licenses/by/4.0/ | We demonstrate background-free imaging and sideband cooling of a single 133Cs
atom via the narrow-line 6S1/2 to 5D5/2 electric quadrupole transition in a
1064 nm optical tweezer. The 5D5/2 state decays through the 6P3/2 state to the
ground state, emitting an 852 nm wavelength photon that allows for
background-free imaging. By encoding both spin and orbital angular momentum
onto the 685 nm excitation light, we achieve background-free fluorescence
histograms with 99.58(3)% fidelity by positioning the atom at the dark center
of a vortex beam. Tuning the tweezer polarization ellipticity realizes a magic
trap for the stretched F = 4, mF = 4 to F' = 6, mF' = 6 cycling transition. We
cool to 5 uK in a 1.1 mK trap and outline a strategy for ground-state cooling.
We compare cooling performance across different sideband regimes, while also
exploring how the orbital angular momentum of structured light controls the
selection rules for quadrupole transitions. These results expand the toolbox
for high-fidelity quantum control and cooling in alkali-atom tweezer arrays.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:50:39 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Blodgett",
"Karl N.",
""
],
[
"Phatak",
"Saumitra S.",
""
],
[
"Chen",
"Meng Raymond",
""
],
[
"Peana",
"David",
""
],
[
"Pritts",
"Claire",
""
],
[
"Hood",
"Jonathan D.",
""
]
] |
2505.10541 | Pengfei Wang | Pengfei Wang, Guohai Xu, Weinong Wang, Junjie Yang, Jie Lou, Yunhua
Xue | Exploring Implicit Visual Misunderstandings in Multimodal Large Language
Models through Attention Analysis | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advancements have enhanced the capability of Multimodal Large Language
Models (MLLMs) to comprehend multi-image information. However, existing
benchmarks primarily evaluate answer correctness, overlooking whether models
genuinely comprehend the visual input. To address this, we define implicit
visual misunderstanding (IVM), where MLLMs provide correct answers without
fully comprehending the visual input. Through our analysis, we decouple the
visual and textual modalities within the causal attention module, revealing
that attention distribution increasingly converges on the image associated with
the correct answer as the network layers deepen. This insight leads to the
introduction of a scale-agnostic metric, \textit{attention accuracy}, and a
novel benchmark for quantifying IVMs. Attention accuracy directly evaluates the
model's visual understanding via internal mechanisms, remaining robust to
positional biases for more reliable assessments. Furthermore, we extend our
approach to finer granularities and demonstrate its effectiveness in unimodal
scenarios, underscoring its versatility and generalizability.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:52:40 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Wang",
"Pengfei",
""
],
[
"Xu",
"Guohai",
""
],
[
"Wang",
"Weinong",
""
],
[
"Yang",
"Junjie",
""
],
[
"Lou",
"Jie",
""
],
[
"Xue",
"Yunhua",
""
]
] |
2505.10542 | Tyler Wilson | Tyler Wilson, Wil Thomason, Zachary Kingston, and Jonathan Gammell | AORRTC: Almost-Surely Asymptotically Optimal Planning with RRT-Connect | 8 pages, 4 figures, 1 table. A video of AORRTC can be found at
https://www.youtube.com/watch?v=j1itxP3KuiM . Information on the
implementation of AORRTC is available at https://robotic-esp.com/code/aorrtc/ | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Finding high-quality solutions quickly is an important objective in motion
planning. This is especially true for high-degree-of-freedom robots.
Satisficing planners have traditionally found feasible solutions quickly but
provide no guarantees on their optimality, while almost-surely asymptotically
optimal (a.s.a.o.) planners have probabilistic guarantees on their convergence
towards an optimal solution but are more computationally expensive.
This paper uses the AO-x meta-algorithm to extend the satisficing RRT-Connect
planner to optimal planning. The resulting Asymptotically Optimal RRT-Connect
(AORRTC) finds initial solutions in similar times as RRT-Connect and uses any
additional planning time to converge towards the optimal solution in an anytime
manner. It is proven to be probabilistically complete and a.s.a.o.
AORRTC was tested with the Panda (7 DoF) and Fetch (8 DoF) robotic arms on
the MotionBenchMaker dataset. These experiments show that AORRTC finds initial
solutions as fast as RRT-Connect and faster than the tested state-of-the-art
a.s.a.o. algorithms while converging to better solutions faster. AORRTC finds
solutions to difficult high-DoF planning problems in milliseconds where the
other a.s.a.o. planners could not consistently find solutions in seconds. This
performance was demonstrated both with and without single instruction/multiple
data (SIMD) acceleration.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:53:11 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Wilson",
"Tyler",
""
],
[
"Thomason",
"Wil",
""
],
[
"Kingston",
"Zachary",
""
],
[
"Gammell",
"Jonathan",
""
]
] |
2505.10543 | Annie Wong | Annie Wong, Thomas B\"ack, Aske Plaat, Niki van Stein and Anna V.
Kononova | Towards a Deeper Understanding of Reasoning Capabilities in Large
Language Models | null | null | null | null | cs.AI cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | While large language models demonstrate impressive performance on static
benchmarks, the true potential of large language models as self-learning and
reasoning agents in dynamic environments remains unclear. This study
systematically evaluates the efficacy of self-reflection, heuristic mutation,
and planning as prompting techniques to test the adaptive capabilities of
agents. We conduct experiments with various open-source language models in
dynamic environments and find that larger models generally outperform smaller
ones, but that strategic prompting can close this performance gap. Second, a
too-long prompt can negatively impact smaller models on basic reactive tasks,
while larger models show more robust behaviour. Third, advanced prompting
techniques primarily benefit smaller models on complex games, but offer less
improvement for already high-performing large language models. Yet, we find
that advanced reasoning methods yield highly variable outcomes: while capable
of significantly improving performance when reasoning and decision-making
align, they also introduce instability and can lead to big performance drops.
Compared to human performance, our findings reveal little evidence of true
emergent reasoning. Instead, large language model performance exhibits
persistent limitations in crucial areas such as planning, reasoning, and
spatial coordination, suggesting that current-generation large language models
still suffer fundamental shortcomings that may not be fully overcome through
self-reflective prompting alone. Reasoning is a multi-faceted task, and while
reasoning methods like Chain of thought improves multi-step reasoning on math
word problems, our findings using dynamic benchmarks highlight important
shortcomings in general reasoning capabilities, indicating a need to move
beyond static benchmarks to capture the complexity of reasoning.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:53:47 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Wong",
"Annie",
""
],
[
"Bäck",
"Thomas",
""
],
[
"Plaat",
"Aske",
""
],
[
"van Stein",
"Niki",
""
],
[
"Kononova",
"Anna V.",
""
]
] |
2505.10544 | Friedrich Simmel | Florian Rothfischer, Lennart J. K. Wei{\ss}, Niccol\`o Tedeschi,
Michael Matthies, Matthias Vogt, Christoph Karfusehr, Alexander Hebel, Petr
\v{S}ulc, Tim Liedl, Enzo Kopperger, Friedrich C. Simmel | High-endurance mechanical switching in a DNA origami snap-through
mechanism | null | null | null | null | physics.app-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Switchable elements are key components of dynamic technological and
biological systems, enabling reversible transitions between well-defined
states. Here, we present a DNA origami-based, mechanically bistable
snap-through mechanism that can be electrically controlled. This nanoscale
switch exhibits long-term stability in both states in the absence of external
stimuli, while achieving millisecond-scale switching times upon application of
an electric field. Individual devices sustain hundreds of thousands of
switching cycles over several hours, offering a powerful platform for
systematically studying the endurance and failure mechanisms of biomolecular
nanoswitches. Functionalization with a gold nanorod further allows
polarization-dependent optical modulation, opening avenues for applications in
plasmonics. This versatile electromechanical interface has potential uses in
molecular information processing, optical nanodevices, and the dynamic control
of chemical reactions.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:53:51 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Rothfischer",
"Florian",
""
],
[
"Weiß",
"Lennart J. K.",
""
],
[
"Tedeschi",
"Niccolò",
""
],
[
"Matthies",
"Michael",
""
],
[
"Vogt",
"Matthias",
""
],
[
"Karfusehr",
"Christoph",
""
],
[
"Hebel",
"Alexander",
""
],
[
"Šulc",
"Petr",
""
],
[
"Liedl",
"Tim",
""
],
[
"Kopperger",
"Enzo",
""
],
[
"Simmel",
"Friedrich C.",
""
]
] |
2505.10545 | Amira Alakhdar | Amira Alakhdar, Barnabas Poczos, Newell Washburn | Pharmacophore-Conditioned Diffusion Model for Ligand-Based De Novo Drug
Design | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Developing bioactive molecules remains a central, time- and cost-heavy
challenge in drug discovery, particularly for novel targets lacking structural
or functional data. Pharmacophore modeling presents an alternative for
capturing the key features required for molecular bioactivity against a
biological target. In this work, we present PharmaDiff, a
pharmacophore-conditioned diffusion model for 3D molecular generation.
PharmaDiff employs a transformer-based architecture to integrate an atom-based
representation of the 3D pharmacophore into the generative process, enabling
the precise generation of 3D molecular graphs that align with predefined
pharmacophore hypotheses. Through comprehensive testing, PharmaDiff
demonstrates superior performance in matching 3D pharmacophore constraints
compared to ligand-based drug design methods. Additionally, it achieves higher
docking scores across a range of proteins in structure-based drug design,
without the need for target protein structures. By integrating pharmacophore
modeling with 3D generative techniques, PharmaDiff offers a powerful and
flexible framework for rational drug design.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:54:29 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Alakhdar",
"Amira",
""
],
[
"Poczos",
"Barnabas",
""
],
[
"Washburn",
"Newell",
""
]
] |
2505.10546 | Shubham Rohal | Shubham Rohal (1), Dong Yoon Lee (1), Phuc Nguyen (2), Shijia Pan (1)
((1) University of California Merced,(2) University of Massachusetts Amherst) | Can On Body Sensing Be Spatial Adaptive? | null | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | Wearable sensors are typically affixed to specific locations on the human
body, and their position remains static, only changing unintentionally due to
motion artifacts. This static configuration introduces significant limitations.
As a result, current systems miss the opportunity to capture dynamic
physiological data from diverse body regions. This research investigates the
potential of developing movable sensors that adaptively reposition themselves
to sample different areas of interest on the body, addressing gaps in spatial
coverage. We designed, developed, and fabricated a 3 x 3 matrix platform to
support moving sensors from one location to another. We validated the
feasibility through simulations on a matrix of up to 9 x 9 locations with up to
16 concurrent sensors and real-world prototype characterization.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:55:08 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Rohal",
"Shubham",
"",
"University of California Merced"
],
[
"Lee",
"Dong Yoon",
"",
"University of California Merced"
],
[
"Nguyen",
"Phuc",
"",
"University of Massachusetts Amherst"
],
[
"Pan",
"Shijia",
"",
"University of California Merced"
]
] |
2505.10547 | Milan Ganai | Milan Ganai, Rohan Sinha, Christopher Agia, Daniel Morton, Marco
Pavone | Real-Time Out-of-Distribution Failure Prevention via Multi-Modal
Reasoning | Website: https://milanganai.github.io/fortress/ | null | null | null | cs.RO cs.AI | http://creativecommons.org/licenses/by/4.0/ | Foundation models can provide robust high-level reasoning on appropriate
safety interventions in hazardous scenarios beyond a robot's training data,
i.e. out-of-distribution (OOD) failures. However, due to the high inference
latency of Large Vision and Language Models, current methods rely on manually
defined intervention policies to enact fallbacks, thereby lacking the ability
to plan generalizable, semantically safe motions. To overcome these challenges
we present FORTRESS, a framework that generates and reasons about semantically
safe fallback strategies in real time to prevent OOD failures. At a low
frequency in nominal operations, FORTRESS uses multi-modal reasoners to
identify goals and anticipate failure modes. When a runtime monitor triggers a
fallback response, FORTRESS rapidly synthesizes plans to fallback goals while
inferring and avoiding semantically unsafe regions in real time. By bridging
open-world, multi-modal reasoning with dynamics-aware planning, we eliminate
the need for hard-coded fallbacks and human safety interventions. FORTRESS
outperforms on-the-fly prompting of slow reasoning models in safety
classification accuracy on synthetic benchmarks and real-world ANYmal robot
data, and further improves system safety and planning success in simulation and
on quadrotor hardware for urban navigation.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:55:28 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Ganai",
"Milan",
""
],
[
"Sinha",
"Rohan",
""
],
[
"Agia",
"Christopher",
""
],
[
"Morton",
"Daniel",
""
],
[
"Pavone",
"Marco",
""
]
] |
2505.10548 | Henrique Soares Assump\c{c}\~ao E Silva | Henrique Assump\c{c}\~ao and Gabriel Coutinho | Semidefinite programming bounds on fractional cut-cover and maximum
2-SAT for highly regular graphs | 13 pages | null | null | null | math.OC math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We use semidefinite programming to bound the fractional cut-cover parameter
of graphs in association schemes in terms of their smallest eigenvalue. We also
extend the equality cases of a primal-dual inequality involving the
Goemans-Williamson semidefinite program, which approximates \textsc{maxcut}, to
graphs in certain coherent configurations. Moreover, we obtain spectral bounds
for \textsc{max 2-sat} when the underlying graphs belong to a symmetric
association scheme by means of a certain semidefinite program used to
approximate quadratic programs, and we further develop this technique in order
to explicitly compute the optimum value of its gauge dual in the case of
distance-regular graphs.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:56:11 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Assumpção",
"Henrique",
""
],
[
"Coutinho",
"Gabriel",
""
]
] |
2505.10549 | Christopher Sutton | Zachary J. L. Bare, CJ Sturgill, Manish Kumar, Iva Milisavljevic,
Hans-Conrad zur Loye, Scott Misture, Morgan Stefik, Christopher Sutton | Computational screening and experimental validation of promising
Wadsley-Roth Niobates | null | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | The growing demand for efficient, high-capacity energy storage systems has
driven extensive research into advanced materials for lithium-ion batteries.
Among the various candidates, Wadsley-Roth (WR) niobates have emerged as a
promising class of materials for fast Li+ storage due to rapid ion diffusion
within their ReO3-like blocks in combination with good electronic conductivity
along the shear planes. Despite the remarkable features of WR phases, there are
presently less than 30 known structures which limits identification of
structure-property relationships for improved performance as well as the
identification of phases with more earth-abundant elements. In this work, we
have dramatically expanded the set of potentially (meta)stable compositions
(with $\Delta$ Hd < 22 meV/atom) to 1301 (out of 3283) through high-throughput
screening with density functional theory (DFT). This large space of compound
was generated through single- and double-site substitution into 10 known
WR-niobate prototypes using 48 elements across the periodic table. To confirm
the structure predictions, we successfully synthesized and validated with X-ray
diffraction a new material, MoWNb24O66. The measured lithium diffusivity in
MoWNb24O66 has a peak value of 1.0x10-16 m2/s at 1.45 V vs. Li/Li+ and achieved
225 mAh/g at 5C. Thus a computationally predicted phase was realized
experimentally with performance exceeding Nb16W5O55, a recent WR benchmark.
Overall, the computational dataset of potentially stable novel compounds and
with one realized that has competitive performance provide a valuable guide for
experimentalists in discovering new durable battery materials.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:56:25 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Bare",
"Zachary J. L.",
""
],
[
"Sturgill",
"CJ",
""
],
[
"Kumar",
"Manish",
""
],
[
"Milisavljevic",
"Iva",
""
],
[
"Loye",
"Hans-Conrad zur",
""
],
[
"Misture",
"Scott",
""
],
[
"Stefik",
"Morgan",
""
],
[
"Sutton",
"Christopher",
""
]
] |
2505.10550 | Milena Horvath | Milena Horvath, Alvise Bastianello, Sudipta Dhar, Rebekka Koch,
Yanliang Guo, Jean-S\'ebastien Caux, Manuele Landini, and Hanns-Christoph
N\"agerl | Observing Bethe strings in an attractive Bose gas far from equilibrium | null | null | null | null | cond-mat.quant-gas cond-mat.stat-mech quant-ph | http://creativecommons.org/licenses/by/4.0/ | Bethe strings are bound states of constituent particles in a variety of
interacting many-body one-dimensional (1D) integrable quantum models relevant
to magnetism, nanophysics, cold atoms and beyond. As emergent fundamental
excitations, they are predicted to collectively reshape observable equilibrium
and dynamical properties. Small individual Bethe strings have recently been
observed in quantum magnets and superconducting qubits. However, creating
states featuring intermixtures of many, including large, strings remains an
outstanding experimental challenge. Here, using nearly integrable ultracold
Bose gases, we realize such intermixtures of Bethe strings out of equilibrium,
by dynamically tuning interactions from repulsive to attractive. We measure the
average binding energy of the strings, revealing the presence of bound states
of more than six particles. We find further evidence for them in the momentum
distribution and in Tan's contact, connected to the correlated density. Our
data quantitatively agree with predictions from generalized hydrodynamics
(GHD). Manipulating intermixtures of Bethe strings opens new avenues for
understanding quantum coherence, nonlinear dynamics and thermalization in
strongly-interacting 1D systems.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:56:49 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Horvath",
"Milena",
""
],
[
"Bastianello",
"Alvise",
""
],
[
"Dhar",
"Sudipta",
""
],
[
"Koch",
"Rebekka",
""
],
[
"Guo",
"Yanliang",
""
],
[
"Caux",
"Jean-Sébastien",
""
],
[
"Landini",
"Manuele",
""
],
[
"Nägerl",
"Hanns-Christoph",
""
]
] |
2505.10551 | Yiwen Liu | Yiwen Liu, Jessica Bader, Jae Myung Kim | Does Feasibility Matter? Understanding the Impact of Feasibility on
Synthetic Training Data | CVPRW 2025 | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the development of photorealistic diffusion models, models trained in
part or fully on synthetic data achieve progressively better results. However,
diffusion models still routinely generate images that would not exist in
reality, such as a dog floating above the ground or with unrealistic texture
artifacts. We define the concept of feasibility as whether attributes in a
synthetic image could realistically exist in the real-world domain; synthetic
images containing attributes that violate this criterion are considered
infeasible. Intuitively, infeasible images are typically considered
out-of-distribution; thus, training on such images is expected to hinder a
model's ability to generalize to real-world data, and they should therefore be
excluded from the training set whenever possible. However, does feasibility
really matter? In this paper, we investigate whether enforcing feasibility is
necessary when generating synthetic training data for CLIP-based classifiers,
focusing on three target attributes: background, color, and texture. We
introduce VariReal, a pipeline that minimally edits a given source image to
include feasible or infeasible attributes given by the textual prompt generated
by a large language model. Our experiments show that feasibility minimally
affects LoRA-fine-tuned CLIP performance, with mostly less than 0.3% difference
in top-1 accuracy across three fine-grained datasets. Also, the attribute
matters on whether the feasible/infeasible images adversarially influence the
classification performance. Finally, mixing feasible and infeasible images in
training datasets does not significantly impact performance compared to using
purely feasible or infeasible datasets.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:57:38 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Liu",
"Yiwen",
""
],
[
"Bader",
"Jessica",
""
],
[
"Kim",
"Jae Myung",
""
]
] |
2505.10552 | Kentaro Barhydt | Kentaro Barhydt, O. Godson Osele, Sreela Kodali, Cosima du Pasquier,
Chase M. Hartquist, H. Harry Asada, Allison M. Okamura | Loop closure grasping: Topological transformations enable strong,
gentle, and versatile grasps | null | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | Grasping mechanisms must both create and subsequently hold grasps that permit
safe and effective object manipulation. Existing mechanisms address the
different functional requirements of grasp creation and grasp holding using a
single morphology, but have yet to achieve the simultaneous strength,
gentleness, and versatility needed for many applications. We present "loop
closure grasping", a class of robotic grasping that addresses these different
functional requirements through topological transformations between open-loop
and closed-loop morphologies. We formalize these morphologies for grasping,
formulate the loop closure grasping method, and present principles and a design
architecture that we implement using soft growing inflated beams, winches, and
clamps. The mechanisms' initial open-loop topology enables versatile grasp
creation via unencumbered tip movement, and closing the loop enables strong and
gentle holding with effectively infinite bending compliance. Loop closure
grasping circumvents the tradeoffs of single-morphology designs, enabling
grasps involving historically challenging objects, environments, and
configurations.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:58:01 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Barhydt",
"Kentaro",
""
],
[
"Osele",
"O. Godson",
""
],
[
"Kodali",
"Sreela",
""
],
[
"Pasquier",
"Cosima du",
""
],
[
"Hartquist",
"Chase M.",
""
],
[
"Asada",
"H. Harry",
""
],
[
"Okamura",
"Allison M.",
""
]
] |
2505.10553 | Jehu Martinez | Jehu Martinez and Andrea Delgado | Flowing Through Hilbert Space: Quantum-Enhanced Generative Models for
Lattice Field Theory | 8 pages, 6 figures, Submitted to QCE Quantum Week 2025 | null | null | null | quant-ph hep-lat | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sampling from high-dimensional and structured probability distributions is a
fundamental challenge in computational physics, particularly in the context of
lattice field theory (LFT), where generating field configurations efficiently
is critical, yet computationally intensive. In this work, we apply a previously
developed hybrid quantum-classical normalizing flow model to explore
quantum-enhanced sampling in such regimes. Our approach embeds parameterized
quantum circuits within a classical normalizing flow architecture, leveraging
amplitude encoding and quantum entanglement to enhance expressivity in the
generative process. The quantum circuit serves as a trainable transformation
within the flow, while classical networks provide adaptive coupling and
compensate for quantum hardware imperfections. This design enables efficient
density estimation and sample generation, potentially reducing the resources
required compared to purely classical methods. While LFT provides a
representative and physically meaningful application for benchmarking, our
focus is on improving the sampling efficiency of generative models through
quantum components. This work contributes toward the development of
quantum-enhanced generative modeling frameworks that address the sampling
bottlenecks encountered in physics and beyond.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:58:16 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Martinez",
"Jehu",
""
],
[
"Delgado",
"Andrea",
""
]
] |
2505.10554 | Zhiyuan Hu | Zhiyuan Hu, Yibo Wang, Hanze Dong, Yuhui Xu, Amrita Saha, Caiming
Xiong, Bryan Hooi, Junnan Li | Beyond 'Aha!': Toward Systematic Meta-Abilities Alignment in Large
Reasoning Models | In Progress | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Large reasoning models (LRMs) already possess a latent capacity for long
chain-of-thought reasoning. Prior work has shown that outcome-based
reinforcement learning (RL) can incidentally elicit advanced reasoning
behaviors such as self-correction, backtracking, and verification phenomena
often referred to as the model's "aha moment". However, the timing and
consistency of these emergent behaviors remain unpredictable and
uncontrollable, limiting the scalability and reliability of LRMs' reasoning
capabilities. To address these limitations, we move beyond reliance on prompts
and coincidental "aha moments". Instead, we explicitly align models with three
meta-abilities: deduction, induction, and abduction, using automatically
generated, self-verifiable tasks. Our three stage-pipeline individual
alignment, parameter-space merging, and domain-specific reinforcement learning,
boosting performance by over 10\% relative to instruction-tuned baselines.
Furthermore, domain-specific RL from the aligned checkpoint yields an
additional 2\% average gain in the performance ceiling across math, coding, and
science benchmarks, demonstrating that explicit meta-ability alignment offers a
scalable and dependable foundation for reasoning. Code is available at:
https://github.com/zhiyuanhubj/Meta-Ability-Alignment
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:58:33 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Hu",
"Zhiyuan",
""
],
[
"Wang",
"Yibo",
""
],
[
"Dong",
"Hanze",
""
],
[
"Xu",
"Yuhui",
""
],
[
"Saha",
"Amrita",
""
],
[
"Xiong",
"Caiming",
""
],
[
"Hooi",
"Bryan",
""
],
[
"Li",
"Junnan",
""
]
] |
2505.10555 | Soumendu Sundar Mukherjee | Soumendu Sundar Mukherjee and Himasish Talukdar | Spectra of contractions of the Gaussian Orthogonal Tensor Ensemble | 45 pages, 1 figure; abstract shortened to meet arXiv requirements | null | null | null | math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this article, we study the spectra of matrix-valued contractions of the
Gaussian Orthogonal Tensor Ensemble (GOTE). Let $\mathcal{G}$ denote a random
tensor of order $r$ and dimension $n$ drawn from the density \[
f(\mathcal{G}) \propto
\exp\bigg(-\frac{1}{2r}\|\mathcal{G}\|^2_{\mathrm{F}}\bigg). \] For $\mathbf{w}
\in \mathbb{S}^{n - 1}$, the unit-sphere in $\mathbb{R}^n$, we consider the
matrix-valued contraction $\mathcal{G} \cdot \mathbf{w}^{\otimes (r - 2)}$ when
both $r$ and $n$ go to infinity such that $r / n \to c \in [0, \infty]$. We
obtain semi-circle bulk-limits in all regimes, generalising the works of
Goulart et al. (2022); Au and Garza-Vargas (2023); Bonnin (2024) in the
fixed-$r$ setting.
We also study the edge-spectrum. We obtain a Baik-Ben Arous-P\'{e}ch\'{e}
phase-transition for the largest and the smallest eigenvalues at $r = 4$,
generalising a result of Mukherjee et al. (2024) in the context of adjacency
matrices of random hypergraphs. We also show that the extreme eigenvectors of
$\mathcal{G} \cdot \mathbf{w}^{\otimes (r - 2)}$ contain non-trivial
information about the contraction direction $\mathbf{w}$.
Finally, we report some results, in the case $r = 4$, on mixed contractions
$\mathcal{G} \cdot \mathbf{u} \otimes \mathbf{v}$, $\mathbf{u}, \mathbf{v} \in
\mathbb{S}^{n - 1}$. While the total variation distance between the joint
distribution of the entries of $\mathcal{G} \cdot \mathbf{u} \otimes
\mathbf{v}$ and that of $\mathcal{G} \cdot \mathbf{u} \otimes \mathbf{u}$ goes
to $0$ when $\|\mathbf{u} - \mathbf{v}\| = o(n^{-1})$, the bulk and the largest
eigenvalues of these two matrices have the same limit profile as long as
$\|\mathbf{u} - \mathbf{v}\| = o(1)$. Furthermore, it turns out that there are
no outlier eigenvalues in the spectrum of $\mathcal{G} \cdot \mathbf{u} \otimes
\mathbf{v}$ when $\langle \mathbf{u}, \mathbf{v} \rangle = o(1)$.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:58:42 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Mukherjee",
"Soumendu Sundar",
""
],
[
"Talukdar",
"Himasish",
""
]
] |
2505.10556 | Claire Heaney | Nazanin Zounemat Kermani, Sadjad Naderi, Claire H. Dilliway, Claire E.
Heaney, Shrreya Behll, Boyang Chen, Hisham Abubakar-Waziri, Alexandra E.
Porter, Marc Chadeau-Hyam, Fangxin Fang, Ian M. Adcock, Kian Fan Chung,
Christopher C. Pain | An AI-driven framework for the prediction of personalised health
response to air pollution | Kermani and Naderi share first authorship. 20 pages, 6 figures and 1
table | null | null | null | cs.LG physics.ao-ph | http://creativecommons.org/licenses/by/4.0/ | Air pollution poses a significant threat to public health, causing or
exacerbating many respiratory and cardiovascular diseases. In addition, climate
change is bringing about more extreme weather events such as wildfires and
heatwaves, which can increase levels of pollution and worsen the effects of
pollution exposure. Recent advances in personal sensing have transformed the
collection of behavioural and physiological data, leading to the potential for
new improvements in healthcare. We wish to capitalise on this data, alongside
new capabilities in AI for making time series predictions, in order to monitor
and predict health outcomes for an individual. Thus, we present a novel
workflow for predicting personalised health responses to pollution by
integrating physiological data from wearable fitness devices with real-time
environmental exposures. The data is collected from various sources in a secure
and ethical manner, and is used to train an AI model to predict individual
health responses to pollution exposure within a cloud-based, modular framework.
We demonstrate that the AI model -- an Adversarial Autoencoder neural network
in this case -- accurately reconstructs time-dependent health signals and
captures nonlinear responses to pollution. Transfer learning is applied using
data from a personal smartwatch, which increases the generalisation abilities
of the AI model and illustrates the adaptability of the approach to real-world,
user-generated data.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:59:07 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Kermani",
"Nazanin Zounemat",
""
],
[
"Naderi",
"Sadjad",
""
],
[
"Dilliway",
"Claire H.",
""
],
[
"Heaney",
"Claire E.",
""
],
[
"Behll",
"Shrreya",
""
],
[
"Chen",
"Boyang",
""
],
[
"Abubakar-Waziri",
"Hisham",
""
],
[
"Porter",
"Alexandra E.",
""
],
[
"Chadeau-Hyam",
"Marc",
""
],
[
"Fang",
"Fangxin",
""
],
[
"Adcock",
"Ian M.",
""
],
[
"Chung",
"Kian Fan",
""
],
[
"Pain",
"Christopher C.",
""
]
] |
2505.10557 | Ke Wang | Ke Wang, Junting Pan, Linda Wei, Aojun Zhou, Weikang Shi, Zimu Lu, Han
Xiao, Yunqiao Yang, Houxing Ren, Mingjie Zhan, Hongsheng Li | MathCoder-VL: Bridging Vision and Code for Enhanced Multimodal
Mathematical Reasoning | Accepted to ACL 2025 Findings | null | null | null | cs.CV cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Natural language image-caption datasets, widely used for training Large
Multimodal Models, mainly focus on natural scenarios and overlook the intricate
details of mathematical figures that are critical for problem-solving,
hindering the advancement of current LMMs in multimodal mathematical reasoning.
To this end, we propose leveraging code as supervision for cross-modal
alignment, since code inherently encodes all information needed to generate
corresponding figures, establishing a precise connection between the two
modalities. Specifically, we co-develop our image-to-code model and dataset
with model-in-the-loop approach, resulting in an image-to-code model,
FigCodifier and ImgCode-8.6M dataset, the largest image-code dataset to date.
Furthermore, we utilize FigCodifier to synthesize novel mathematical figures
and then construct MM-MathInstruct-3M, a high-quality multimodal math
instruction fine-tuning dataset. Finally, we present MathCoder-VL, trained with
ImgCode-8.6M for cross-modal alignment and subsequently fine-tuned on
MM-MathInstruct-3M for multimodal math problem solving. Our model achieves a
new open-source SOTA across all six metrics. Notably, it surpasses GPT-4o and
Claude 3.5 Sonnet in the geometry problem-solving subset of MathVista,
achieving improvements of 8.9% and 9.2%. The dataset and models will be
released at https://github.com/mathllm/MathCoder.
| [
{
"version": "v1",
"created": "Thu, 15 May 2025 17:59:21 GMT"
}
] | 2025-05-16T00:00:00 | [
[
"Wang",
"Ke",
""
],
[
"Pan",
"Junting",
""
],
[
"Wei",
"Linda",
""
],
[
"Zhou",
"Aojun",
""
],
[
"Shi",
"Weikang",
""
],
[
"Lu",
"Zimu",
""
],
[
"Xiao",
"Han",
""
],
[
"Yang",
"Yunqiao",
""
],
[
"Ren",
"Houxing",
""
],
[
"Zhan",
"Mingjie",
""
],
[
"Li",
"Hongsheng",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.