id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2505.22192 | Yue Cui | Yue Cui, Liuyi Yao, Zitao Li, Yaliang Li, Bolin Ding, Xiaofang Zhou | Efficient Leave-one-out Approximation in LLM Multi-agent Debate Based on Introspection | null | null | null | null | cs.MA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multi-agent systems based on large language models (LLMs) advance automatic task completion in various fields, where debate is a common cooperation form for agents to solve complicated problems with reasoning and cross-review to solidify answers. Assessing the individual contributions of agents within these debates is crucial for system refinement and outcome reliability. Traditional leave-one-out (LOO) method offers a clear framework for evaluating each agent's role but face challenges in LLM-based systems due to high computational costs and associated financial implications. This paper presents introspective-leave-one-out (IntrospecLOO), a simple yet effective prompting for approximation of LOO in LLM-powered multi-agent debates. IntrospecLOO introduces an additional querying round after standard debates, prompting agents to update their answers while ignoring responses from a designated agent. This strategy effectively isolates and gauges each participant's influence at a reduced query complexity compared to the original LOO approaches. Validation through experiments on three benchmark datasets confirms the effectiveness of IntrospecLOO. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:08:31 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Cui",
"Yue",
""
],
[
"Yao",
"Liuyi",
""
],
[
"Li",
"Zitao",
""
],
[
"Li",
"Yaliang",
""
],
[
"Ding",
"Bolin",
""
],
[
"Zhou",
"Xiaofang",
""
]
] |
2505.22193 | Marco Parigi | Marco Parigi, Stefano Martina, Francesco Aldo Venturelli and Filippo Caruso | Physics-inspired Generative AI models via real hardware-based noisy quantum diffusion | 17 pages, 9 figures. Supplementary materials: 2 pages, 2 figures | null | null | null | quant-ph cond-mat.dis-nn cs.AI cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Quantum Diffusion Models (QDMs) are an emerging paradigm in Generative AI that aims to use quantum properties to improve the performances of their classical counterparts. However, existing algorithms are not easily scalable due to the limitations of near-term quantum devices. Following our previous work on QDMs, here we propose and implement two physics-inspired protocols. In the first, we use the formalism of quantum stochastic walks, showing that a specific interplay of quantum and classical dynamics in the forward process produces statistically more robust models generating sets of MNIST images with lower Fr\'echet Inception Distance (FID) than using totally classical dynamics. In the second approach, we realize an algorithm to generate images by exploiting the intrinsic noise of real IBM quantum hardware with only four qubits. Our work could be a starting point to pave the way for new scenarios for large-scale algorithms in quantum Generative AI, where quantum noise is neither mitigated nor corrected, but instead exploited as a useful resource. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:11:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Parigi",
"Marco",
""
],
[
"Martina",
"Stefano",
""
],
[
"Venturelli",
"Francesco Aldo",
""
],
[
"Caruso",
"Filippo",
""
]
] |
2505.22194 | Can Xiao | Can Xiao, Jianyi Cheng, Aaron Zhao | Refining Datapath for Microscaling ViTs | Accepted at FPL'2025 | null | null | null | cs.AR | http://creativecommons.org/licenses/by/4.0/ | Vision Transformers (ViTs) leverage the transformer architecture to effectively capture global context, demonstrating strong performance in computer vision tasks. A major challenge in ViT hardware acceleration is that the model family contains complex arithmetic operations that are sensitive to model accuracy, such as the Softmax and LayerNorm operations, which cannot be mapped onto efficient hardware with low precision. Existing methods only exploit parallelism in the matrix multiplication operations of the model on hardware and keep these complex operations on the CPU. This results in suboptimal performance due to the communication overhead between the CPU and accelerator. Can new data formats solve this problem?
In this work, we present the first ViT accelerator that maps all operations of the ViT models onto FPGAs. We exploit a new arithmetic format named Microscaling Integer (MXInt) for datapath designs and evaluate how different design choices can be made to trade off accuracy, hardware performance, and hardware utilization. Our contributions are twofold. First, we quantize ViTs using the MXInt format, achieving both high area efficiency and accuracy. Second, we propose MXInt-specific hardware optimization that map these complex arithmetic operations into custom hardware. Within 1\% accuracy loss, our method achieves at least 93$\times$ speedup compared to Float16 and at least 1.9$\times$ speedup compared to related work. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:15:37 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Xiao",
"Can",
""
],
[
"Cheng",
"Jianyi",
""
],
[
"Zhao",
"Aaron",
""
]
] |
2505.22195 | Guangwei Gao | Guoan Xu, Wenfeng Huang, Wenjing Jia, Jiamao Li, Guangwei Gao, Guo-Jun Qi | S2AFormer: Strip Self-Attention for Efficient Vision Transformer | 12 pages, 6 figures, 8 tables | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Vision Transformer (ViT) has made significant advancements in computer vision, thanks to its token mixer's sophisticated ability to capture global dependencies between all tokens. However, the quadratic growth in computational demands as the number of tokens increases limits its practical efficiency. Although recent methods have combined the strengths of convolutions and self-attention to achieve better trade-offs, the expensive pairwise token affinity and complex matrix operations inherent in self-attention remain a bottleneck. To address this challenge, we propose S2AFormer, an efficient Vision Transformer architecture featuring novel Strip Self-Attention (SSA). We design simple yet effective Hybrid Perception Blocks (HPBs) to effectively integrate the local perception capabilities of CNNs with the global context modeling of Transformer's attention mechanisms. A key innovation of SSA lies in its reducing the spatial dimensions of $K$ and $V$ while compressing the channel dimensions of $Q$ and $K$. This design significantly reduces computational overhead while preserving accuracy, striking an optimal balance between efficiency and effectiveness. We evaluate the robustness and efficiency of S2AFormer through extensive experiments on multiple vision benchmarks, including ImageNet-1k for image classification, ADE20k for semantic segmentation, and COCO for object detection and instance segmentation. Results demonstrate that S2AFormer achieves significant accuracy gains with superior efficiency in both GPU and non-GPU environments, making it a strong candidate for efficient vision Transformers. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:17:23 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Xu",
"Guoan",
""
],
[
"Huang",
"Wenfeng",
""
],
[
"Jia",
"Wenjing",
""
],
[
"Li",
"Jiamao",
""
],
[
"Gao",
"Guangwei",
""
],
[
"Qi",
"Guo-Jun",
""
]
] |
2505.22196 | Jingyi Cui | Jingyi Cui, Hongwei Wen, Yisen Wang | An Augmentation-Aware Theory for Self-Supervised Contrastive Learning | Accepted to ICML2025 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-supervised contrastive learning has emerged as a powerful tool in machine learning and computer vision to learn meaningful representations from unlabeled data. Meanwhile, its empirical success has encouraged many theoretical studies to reveal the learning mechanisms. However, in the existing theoretical research, the role of data augmentation is still under-exploited, especially the effects of specific augmentation types. To fill in the blank, we for the first time propose an augmentation-aware error bound for self-supervised contrastive learning, showing that the supervised risk is bounded not only by the unsupervised risk, but also explicitly by a trade-off induced by data augmentation. Then, under a novel semantic label assumption, we discuss how certain augmentation methods affect the error bound. Lastly, we conduct both pixel- and representation-level experiments to verify our proposed theoretical results. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:18:20 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Cui",
"Jingyi",
""
],
[
"Wen",
"Hongwei",
""
],
[
"Wang",
"Yisen",
""
]
] |
2505.22197 | Giuseppe Antonio Zampogna | Kevin Wittkowski, Pier Giuseppe Ledda, Edoardo Carlo Giordano, Fran\c{c}ois Gallaire, Giuseppe Antonio Zampogna | Effective flows across diffusio-phoretic membranes | null | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | Flows enabled by phoretic mechanisms are of significant interest in several biological and biomedical processes, such as bacterial motion and targeted drug delivery. Here, we develop a homogenization-based macroscopic boundary condition which describes the effective flow across a diffusiophoretic microstructured membrane, where the interaction between the membrane walls and the solute particles is modeled via a potential-approach. We consider two cases where potential variations occur (i) at the pore scale and (ii) only in the close vicinity of the boundary, enabling a simplified version of the macroscopic flow description, in the latter case. Chemical interactions at the microscale are rigorously upscaled to macroscopic phoretic solvent velocity and solute flux contributions, and added to the classical permeability and diffusivity properties of the membrane. These properties stem from the solution of Stokes-advection-diffusion problems at the microscale, some of them forced by an interaction potential term. Eventually, we show an application of the macroscopic model to develop minimal phoretic pumps, showcasing its suitability for efficient design and optimization procedures. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:18:46 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wittkowski",
"Kevin",
""
],
[
"Ledda",
"Pier Giuseppe",
""
],
[
"Giordano",
"Edoardo Carlo",
""
],
[
"Gallaire",
"François",
""
],
[
"Zampogna",
"Giuseppe Antonio",
""
]
] |
2505.22198 | Thomas Elsaesser | Matthias Runge, Michael Woerner, Denys I. Bondar, Thomas Elsaesser | Solvated electrons in polar liquids as epsilon-near-zero materials tunable in the terahertz frequency range | null | Phys. Rev. Lett. 134, 056901 (2025) | 10.1103/PhysRevLett.134.056901 | null | physics.chem-ph cond-mat.mtrl-sci | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Electrons in polar liquids give rise to a polaron resonance at a terahertz (THz) frequency \nu_0 depending on electron concentration. The impact of this resonance on light propagation is studied in experiments, where a femtosecond pump pulse generates electrons via multiphoton ionization and a THz probe pulse propagated through the excited sample is detected in a phase-resolved way. We observe a behavior characteristic for epsilon-near-zero (ENZ) materials with strongly modified phase and group velocities around \nu_0, and a broadening of the THz pulse envelope below \nu_0. Calculations based on a local-field approach reproduce the ENZ behavior. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:22:06 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Runge",
"Matthias",
""
],
[
"Woerner",
"Michael",
""
],
[
"Bondar",
"Denys I.",
""
],
[
"Elsaesser",
"Thomas",
""
]
] |
2505.22199 | Xinyue Hu | Xinyue Hu, Zhibin Duan, Bo Chen, Mingyuan Zhou | Enhancing Uncertainty Estimation and Interpretability via Bayesian Non-negative Decision Layer | Accepted by The Thirteenth International Conference on Learning Representations (ICLR 2025) | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Although deep neural networks have demonstrated significant success due to their powerful expressiveness, most models struggle to meet practical requirements for uncertainty estimation. Concurrently, the entangled nature of deep neural networks leads to a multifaceted problem, where various localized explanation techniques reveal that multiple unrelated features influence the decisions, thereby undermining interpretability. To address these challenges, we develop a Bayesian Non-negative Decision Layer (BNDL), which reformulates deep neural networks as a conditional Bayesian non-negative factor analysis. By leveraging stochastic latent variables, the BNDL can model complex dependencies and provide robust uncertainty estimation. Moreover, the sparsity and non-negativity of the latent variables encourage the model to learn disentangled representations and decision layers, thereby improving interpretability. We also offer theoretical guarantees that BNDL can achieve effective disentangled learning. In addition, we developed a corresponding variational inference method utilizing a Weibull variational inference network to approximate the posterior distribution of the latent variables. Our experimental results demonstrate that with enhanced disentanglement capabilities, BNDL not only improves the model's accuracy but also provides reliable uncertainty estimation and improved interpretability. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:23:34 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hu",
"Xinyue",
""
],
[
"Duan",
"Zhibin",
""
],
[
"Chen",
"Bo",
""
],
[
"Zhou",
"Mingyuan",
""
]
] |
2505.22200 | Darshana Saravanan | Darshana Saravanan, Makarand Tapaswi, Vineet Gandhi | Investigating Mechanisms for In-Context Vision Language Binding | Accepted to MIV at CVPRW 2025 (Oral) | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | To understand a prompt, Vision-Language models (VLMs) must perceive the image, comprehend the text, and build associations within and across both modalities. For instance, given an 'image of a red toy car', the model should associate this image to phrases like 'car', 'red toy', 'red object', etc. Feng and Steinhardt propose the Binding ID mechanism in LLMs, suggesting that the entity and its corresponding attribute tokens share a Binding ID in the model activations. We investigate this for image-text binding in VLMs using a synthetic dataset and task that requires models to associate 3D objects in an image with their descriptions in the text. Our experiments demonstrate that VLMs assign a distinct Binding ID to an object's image tokens and its textual references, enabling in-context association. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:25:43 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Saravanan",
"Darshana",
""
],
[
"Tapaswi",
"Makarand",
""
],
[
"Gandhi",
"Vineet",
""
]
] |
2505.22201 | Alexander Dimoff | A.J. Dimoff, R.J. Stancliffe, C.J. Hansen, R.M. Seeburger, H. Taylor | Modeling the progenitors of low-mass post-accretion binaries | null | null | null | null | astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | About half of the mass of all heavy elements with mass number A > 90 is formed through the slow neutron capture process (s-process), occurring in evolved asymptotic giant branch (AGB) stars with masses ~1-6 $\rm{M_{\odot}}$. The s-process can be studied by modeling the accretion of material from AGB stars onto binary barium (Ba), CH, and carbon-enhanced metal-poor (CEMP)-s stars. Comparing observationally derived surface parameters and 1D-LTE abundance patterns of s-process elements to theoretical binary accretion models, we aim to understand the formation of post-accretion systems. We explore the extent of dilution of the accreted material and describe the impact of convective mixing on the observed surface abundances. We compute a new grid of 2700 accretion models for low-mass post-accretion systems. A maximum-likelihood comparison determines the best fit models for observational samples of Ba, CH, and CEMP-s stars. We find consistent AGB donor masses in the mass range of 2-3 $\rm{M_{\odot}}$ across our sample of post-accretion binaries. We find the formation scenario for weak Ba stars is an AGB star transferring a moderate amount of mass ($\leq$0.5 $\rm{M_{\odot}}$) resulting in a ~2.0-2.5 $\rm{M_{\odot}}$ star. The strong Ba stars are best fit with lower final masses ~1.0-2.0 $\rm{M_{\odot}}$, and significant accreted mass ($\geq$0.5 $\rm{M_{\odot}}$). The CH and CEMP-s stars display lower final masses (~1.0 $\rm{M_{\odot}}$) and small amounts of transferred material (~0.1 $\rm{M_{\odot}}$). We find that Ba stars generally accrete more material than CEMP-s and CH stars. We also find that strong Ba stars must accrete more than 0.50 $\rm{M_{\odot}}$ to explain their abundance patterns, and in this limit we are unable to reproduce the observed mass distribution of strong Ba stars. The mass distributions of the weak Ba stars, CEMP-s, and CH stars are well reproduced in our modeling. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:27:54 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Dimoff",
"A. J.",
""
],
[
"Stancliffe",
"R. J.",
""
],
[
"Hansen",
"C. J.",
""
],
[
"Seeburger",
"R. M.",
""
],
[
"Taylor",
"H.",
""
]
] |
2505.22202 | Hyeonbin Hwang | Hyeonbin Hwang, Byeongguk Jeon, Seungone Kim, Jiyeon Kim, Hoyeon Chang, Sohee Yang, Seungpil Won, Dohaeng Lee, Youbin Ahn, Minjoon Seo | Let's Predict Sentence by Sentence | Work In Progress | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Autoregressive language models (LMs) generate one token at a time, yet human reasoning operates over higher-level abstractions - sentences, propositions, and concepts. This contrast raises a central question- Can LMs likewise learn to reason over structured semantic units rather than raw token sequences? In this work, we investigate whether pretrained LMs can be lifted into such abstract reasoning spaces by building on their learned representations. We present a framework that adapts a pretrained token-level LM to operate in sentence space by autoregressively predicting continuous embeddings of next sentences. We explore two embedding paradigms inspired by classical representation learning: 1) semantic embeddings, learned via autoencoding to preserve surface meaning; and 2) contextual embeddings, trained via next-sentence prediction to encode anticipatory structure. We evaluate both under two inference regimes: Discretized, which decodes each predicted embedding into text before re-encoding; and Continuous, which reasons entirely in embedding space for improved efficiency. Across four domains - mathematics, logic, commonsense, and planning - contextual embeddings under continuous inference show competitive performance with Chain-of-Thought (CoT) while reducing inference-time FLOPs on average by half. We also present early signs of scalability and modular adaptation. Finally, to visualize latent trajectories, we introduce SentenceLens, a diagnostic tool that decodes intermediate model states into interpretable sentences. Together, our results indicate that pretrained LMs can effectively transition to abstract, structured reasoning within latent embedding spaces. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:28:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hwang",
"Hyeonbin",
""
],
[
"Jeon",
"Byeongguk",
""
],
[
"Kim",
"Seungone",
""
],
[
"Kim",
"Jiyeon",
""
],
[
"Chang",
"Hoyeon",
""
],
[
"Yang",
"Sohee",
""
],
[
"Won",
"Seungpil",
""
],
[
"Lee",
"Dohaeng",
""
],
[
"Ahn",
"Youbin",
""
],
[
"Seo",
"Minjoon",
""
]
] |
2505.22203 | Yuzhen Huang | Yuzhen Huang, Weihao Zeng, Xingshan Zeng, Qi Zhu, Junxian He | Pitfalls of Rule- and Model-based Verifiers -- A Case Study on Mathematical Reasoning | null | null | null | null | cs.LG cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Trustworthy verifiers are essential for the success of reinforcement learning with verifiable reward (RLVR), which is the core methodology behind various large reasoning models such as DeepSeek-R1. In complex domains like mathematical reasoning, rule-based verifiers have been widely adopted in previous works to train strong reasoning models. However, the reliability of these verifiers and their impact on the RL training process remain poorly understood. In this work, we take mathematical reasoning as a case study and conduct a comprehensive analysis of various verifiers in both static evaluation and RL training scenarios. First, we find that current open-source rule-based verifiers often fail to recognize equivalent answers presented in different formats across multiple commonly used mathematical datasets, resulting in non-negligible false negative rates. This limitation adversely affects RL training performance and becomes more pronounced as the policy model gets stronger. Subsequently, we investigate model-based verifiers as a potential solution to address these limitations. While the static evaluation shows that model-based verifiers achieve significantly higher verification accuracy, further analysis and RL training results imply that they are highly susceptible to hacking, where they misclassify certain patterns in responses as correct (i.e., false positives). This vulnerability is exploited during policy model optimization, leading to artificially inflated rewards. Our findings underscore the unique risks inherent to both rule-based and model-based verifiers, aiming to offer valuable insights to develop more robust reward systems in reinforcement learning. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:28:41 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Huang",
"Yuzhen",
""
],
[
"Zeng",
"Weihao",
""
],
[
"Zeng",
"Xingshan",
""
],
[
"Zhu",
"Qi",
""
],
[
"He",
"Junxian",
""
]
] |
2505.22204 | Ashadul Halder | Diptarko Mukherjee, Ashadul Halder, Debasish Majumdar, Abhijit Bandyopadhyay | IceCube PeV neutrinos from heavy dark matter decay with 12 years HESE data | 8 pages LaTeX, 4 figures | null | null | null | hep-ph astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The decay of superheavy dark matter from the early universe may undergo decay via QCD cascades and electroweak cascade to produce neutrinos as one of the decay products. We consider the neutrino events in and around PeV region reported by IceCube collaboration are due to the decay of such heavy dark matter. The neutrino spectrum could be from the decay processes via hadronic decay modes and/or leptonic decay modes. Using the numerical evolution of QCD cascades as well as electroweak corrections where use has been made of DGLAP equations, the neutrino fluxes from the heavy dark matter decay have been computed. The mass of the decaying superheavy dark matter and its decay lifetime have then been estimated from a $\chi^2$ analysis of the IceCube 12-year data. The fractional contribution ($f_{\rm lep}$) of the leptonic decay channel in such a decay process is also estimated from the same $\chi^2$ analyses. It is seen that to explain the IceCube 12-year ultrahigh energy (UHE) events the mass of a decaying superheavy dark matter would be $\sim9.4\times 10^6$ GeV and decay time $\tau \simeq 4.2 \times 10^{28}$ second. It is also found that the lepton channel contribution is very small, $f_{\rm lep} \sim 0.001$. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:30:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mukherjee",
"Diptarko",
""
],
[
"Halder",
"Ashadul",
""
],
[
"Majumdar",
"Debasish",
""
],
[
"Bandyopadhyay",
"Abhijit",
""
]
] |
2505.22205 | Patrick De Laverny | Patrick de Laverny, Roxanne Ligi, Aur\'elien Crida, Alejandra Recio-Blanco, Pedro A. Palicio | The Gaia spectroscopic catalogue of exoplanets and host stars | A&A in press | null | null | null | astro-ph.EP astro-ph.IM astro-ph.SR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Complete, accurate, and precise catalogues of exoplanet host star (EHS) properties are essential to deriving high-quality exoplanet parameters. This paper aims at homogeneously parameterising EHS and their exoplanets, using Gaia and GSP-spec data. For 2573 EHS, we computed their L*, R*, and M*, with no prior assumption from stellar evolution models. Their Galactic positions, kinematic and orbital properties were derived. We re-scaled Mp and Rp of 3556 exoplanets, fully consistently with the stellar data. These new stellar Teff, L*, and R* are in rather good agreement with literature values but are more precise. In particular, R* are derived with typically less than 3% uncertainty; this reduces the uncertainty on Rp significantly and allows for a finer analysis of the decrease in the number of planets around the evaporation valley. Larger differences, however, were found for M*. We note that the EHS population is rather diverse in terms of the chemical and Galactic properties, although they are all found in the Solar vicinity, close to the Local spiral arm. Most EHS belong to the thin disc, but some older thick disc and halo members have also been identified. For the less massive planets, the average Rp increases with the metallicity of the host star. For giant planets, a dichotomy between dense and inflated planets is found. Denser planets tend to be more massive as the metallicity of the host star increases, while inflated planets are more massive for less metallic hosts. If confirmed, this bimodality implies that the diversity of giant exoplanets depends on their Galactic birth locus, with dense giant planets being more numerous than inflated ones when [M/H]>~1.5 times Solar, as in the central Milky Way regions. The Gaia spectroscopic catalogue of exoplanets and their host stars is large, homogeneous, and precise. It would be a useful added-value for planetary studies. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:33:09 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"de Laverny",
"Patrick",
""
],
[
"Ligi",
"Roxanne",
""
],
[
"Crida",
"Aurélien",
""
],
[
"Recio-Blanco",
"Alejandra",
""
],
[
"Palicio",
"Pedro A.",
""
]
] |
2505.22206 | Manuel \'Ubeda-Flores | Enrique de Amo, David Garc\'ia-Fern\'andez, Manuel \'Ubeda-Flores | Directional $\rho$-coefficients | 1 figure | null | null | null | math.ST stat.TH | http://creativecommons.org/licenses/by/4.0/ | In this paper we obtain advances for the concept of directional $\rho$-coefficients, originally defined for the trivariate case in [Nelsen, R.B., \'Ubeda-Flores, M. (2011). Directional dependence in multivariate distributions. Ann. Inst. Stat. Math 64, 677-685] by extending it to encompass arbitrary dimensions and directions in multivariate space. We provide a generalized definition and establish its fundamental properties. Moreover, we resolve a conjecture from the aforementioned work by proving a more general result applicable to any dimension, correcting a result in [Garc\'ia, J.E., Gonz\'alez-L\'opez, V.A., Nelsen, R.B. (2013). A new index to measure positive dependence in trivariate distributions. J. Multivariate Anal. 115, 481-495] an erratum in the current literature. Our findings contribute to a deeper understanding of multivariate dependence and association, offering novel tools for detecting directional dependencies in high-dimensional settings. Finally, we introduce nonparametric estimators, based on ranks, for estimating directional $\rho$-coefficients from a sample. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:33:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"de Amo",
"Enrique",
""
],
[
"García-Fernández",
"David",
""
],
[
"Úbeda-Flores",
"Manuel",
""
]
] |
2505.22207 | Akshay Mahajan | Akshay Mahajan and Awadhesh Narayan | Modulation of Polarization and Metallicity in Janus Sliding Ferroelectrics | null | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sliding ferroelectricity is emerging as a distinct and promising mechanism for realizing ferroelectricity in low-dimensional systems, offering new design principles beyond the conventional ferroelectric mechanism. Further, the coexistence of the out-of-plane polarization with in-plane conductivity induced by electrostatic charge doping makes these systems strong candidates for realizing ferroelectric metals. Using density functional theory calculations, we analyze the transition metal dichalcogenides (TMDs) based Janus sliding ferroelectric bilayers XMY (M = Mo, W; X, Y = S, Se, Te; X $\neq$ Y). In addition to exhibiting switchable interlayer polarization, Janus sliding ferroelectrics possess an intrinsic electric field within each monolayer, arising from the electronegativity difference between the chalcogen atoms. We discover that the intrinsic electric field of the monolayers can be used to modulate the interlayer ferroelectric polarization and the electronic band structure. We identify the decrease in the interlayer distance due to a particular stacking of the Janus bilayers as a major contributor to increasing polarization and reducing the bandgap. The direction of the intrinsic electric field within the Janus monolayers plays a significant role in the modulation of layer-wise contribution in the valence and conduction bands, which influences the polarization reduction due to extrinsic charge dopants. Extending this concept to Janus trilayers, we observe further enhancement in polarization and additional bandgap reduction compared to their bilayer counterparts. These results highlight the tunability of TMD-based Janus sliding ferroelectrics and suggest a pathway for designing low bandgap ferroelectrics and potential ferroelectric metals. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:34:16 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mahajan",
"Akshay",
""
],
[
"Narayan",
"Awadhesh",
""
]
] |
2505.22208 | Yosuke Oyama | Yosuke Oyama, Yusuke Majima, Eiji Ohta, Yasufumi Sakai | LaMM: Semi-Supervised Pre-Training of Large-Scale Materials Models | 24 pages, 9 figures | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Neural network potentials (NNPs) are crucial for accelerating computational materials science by surrogating density functional theory (DFT) calculations. Improving their accuracy is possible through pre-training and fine-tuning, where an NNP model is first pre-trained on a large-scale dataset and then fine-tuned on a smaller target dataset. However, this approach is computationally expensive, mainly due to the cost of DFT-based dataset labeling and load imbalances during large-scale pre-training. To address this, we propose LaMM, a semi-supervised pre-training method incorporating improved denoising self-supervised learning and a load-balancing algorithm for efficient multi-node training. We demonstrate that our approach effectively leverages a large-scale dataset of $\sim$300 million semi-labeled samples to train a single NNP model, resulting in improved fine-tuning performance in terms of both speed and accuracy. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:36:49 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Oyama",
"Yosuke",
""
],
[
"Majima",
"Yusuke",
""
],
[
"Ohta",
"Eiji",
""
],
[
"Sakai",
"Yasufumi",
""
]
] |
2505.22209 | Naomi Kombol | Naomi Kombol, Ivan Martinovi\'c, Sini\v{s}a \v{S}egvi\'c | A Survey on Training-free Open-Vocabulary Semantic Segmentation | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Semantic segmentation is one of the most fundamental tasks in image understanding with a long history of research, and subsequently a myriad of different approaches. Traditional methods strive to train models up from scratch, requiring vast amounts of computational resources and training data. In the advent of moving to open-vocabulary semantic segmentation, which asks models to classify beyond learned categories, large quantities of finely annotated data would be prohibitively expensive. Researchers have instead turned to training-free methods where they leverage existing models made for tasks where data is more easily acquired. Specifically, this survey will cover the history, nuance, idea development and the state-of-the-art in training-free open-vocabulary semantic segmentation that leverages existing multi-modal classification models. We will first give a preliminary on the task definition followed by an overview of popular model archetypes and then spotlight over 30 approaches split into broader research branches: purely CLIP-based, those leveraging auxiliary visual foundation models and ones relying on generative methods. Subsequently, we will discuss the limitations and potential problems of current research, as well as provide some underexplored ideas for future study. We believe this survey will serve as a good onboarding read to new researchers and spark increased interest in the area. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:37:52 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kombol",
"Naomi",
""
],
[
"Martinović",
"Ivan",
""
],
[
"Šegvić",
"Siniša",
""
]
] |
2505.22210 | Nan Liu | Nan Liu and Ran Wang | Soliton resolution for the coupled complex short pulse equation | null | null | null | null | nlin.SI math-ph math.AP math.MP | http://creativecommons.org/licenses/by/4.0/ | We address the long-time asymptotics of the solution to the Cauchy problem of ccSP (coupled complex short pulse) equation on the line for decaying initial data that can support solitons. The ccSP system describes ultra-short pulse propagation in optical fibers, which is a completely integrable system and posses a $4\times4$ matrix Wadati--Konno--Ichikawa type Lax pair. Based on the $\bar{\partial}$-generalization of the Deift--Zhou steepest descent method, we obtain the long-time asymptotic approximations of the solution in two kinds of space-time regions under a new scale $(\zeta,t)$. The solution of the ccSP equation decays as a speed of $O(t^{-1})$ in the region $\zeta/t>\varepsilon$ with any $\varepsilon>0$; while in the region $\zeta/t<-\varepsilon$, the solution is depicted by the form of a multi-self-symmetric soliton/composite breather and $t^{-1/2}$ order term arises from self-symmetric soliton/composite breather-radiation interactions as well as an residual error order $O(t^{-1}\ln t)$. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:38:15 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Liu",
"Nan",
""
],
[
"Wang",
"Ran",
""
]
] |
2505.22211 | The Tien Mai | The Tien Mai | Handling bounded response in high dimensions: a Horseshoe prior Bayesian Beta regression approach | null | null | null | null | stat.ME math.ST stat.ML stat.TH | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Bounded continuous responses -- such as proportions -- arise frequently in diverse scientific fields including climatology, biostatistics, and finance. Beta regression is a widely adopted framework for modeling such data, due to the flexibility of the Beta distribution over the unit interval. While Bayesian extensions of Beta regression have shown promise, existing methods are limited to low-dimensional settings and lack theoretical guarantees. In this work, we propose a novel Bayesian approach for high-dimensional sparse Beta regression framework that employs a tempered posterior. Our method incorporates the Horseshoe prior for effective shrinkage and variable selection. Most notable, we propose a novel Gibbs sampling algorithm using P\'olya-Gamma augmentation for efficient inference in Beta regression model. We also provide the first theoretical results establishing posterior consistency and convergence rates for Bayesian Beta regression. Through extensive simulation studies in both low- and high-dimensional scenarios, we demonstrate that our approach outperforms existing alternatives, offering improved estimation accuracy and model interpretability.
Our method is implemented in the R package ``betaregbayes" available on Github. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:39:05 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mai",
"The Tien",
""
]
] |
2505.22212 | Tung Anh Vu | Christoph Hunkenschr\"oder, Martin Kouteck\'y, Asaf Levin, Tung Anh Vu | (Near)-Optimal Algorithms for Sparse Separable Convex Integer Programs | 28 pages, will appear at IPCO 2025 | null | null | null | cs.DS math.OC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We study the general integer programming (IP) problem of optimizing a separable convex function over the integer points of a polytope: $\min \{f(\mathbf{x}) \mid A\mathbf{x} = \mathbf{b}, \, \mathbf{l} \leq \mathbf{x} \leq \mathbf{u}, \, \mathbf{x} \in \mathbb{Z}^n\}$. The number of variables $n$ is a variable part of the input, and we consider the regime where the constraint matrix $A$ has small coefficients $\|A\|_\infty$ and small primal or dual treedepth $\mathrm{td}_P(A)$ or $\mathrm{td}_D(A)$, respectively. Equivalently, we consider block-structured matrices, in particular $n$-fold, tree-fold, $2$-stage and multi-stage matrices.
We ask about the possibility of near-linear time algorithms in the general case of (non-linear) separable convex functions. The techniques of previous works for the linear case are inherently limited to it; in fact, no strongly-polynomial algorithm may exist due to a simple unconditional information-theoretic lower bound of $n \log \|\mathbf{u}-\mathbf{l}\|_\infty$, where $\mathbf{l}, \mathbf{u}$ are the vectors of lower and upper bounds. Our first result is that with parameters $\mathrm{td}_P(A)$ and $\|A\|_\infty$, this lower bound can be matched (up to dependency on the parameters). Second, with parameters $\mathrm{td}_D(A)$ and $\|A\|_\infty$, the situation is more involved, and we design an algorithm with time complexity $g(\mathrm{td}_D(A), \|A\|_\infty) n \log n \log \|\mathbf{u}-\mathbf{l}\|_\infty$ where $g$ is some computable function. We conjecture that a stronger lower bound is possible in this regime, and our algorithm is in fact optimal.
Our algorithms combine ideas from scaling, proximity, and sensitivity of integer programs, together with a new dynamic data structure. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:39:56 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hunkenschröder",
"Christoph",
""
],
[
"Koutecký",
"Martin",
""
],
[
"Levin",
"Asaf",
""
],
[
"Vu",
"Tung Anh",
""
]
] |
2505.22213 | Marton Hajdu | M\'arton Hajdu and Laura Kov\'acs and Andrei Voronkov | Partial Redundancy in Saturation | null | null | null | null | cs.LO | http://creativecommons.org/licenses/by/4.0/ | Redundancy elimination is one of the crucial ingredients of efficient saturation-based proof search. We improve redundancy elimination by introducing a new notion of redundancy, based on partial clauses and redundancy formulas, which is more powerful than the standard notion: there are both clauses and inferences that are redundant when we use our notions and not redundant when we use standard notions. In a way, our notion blurs the distinction between redundancy at the level of inferences and redundancy at the level of clauses. We present a superposition calculus PaRC on partial clauses. Our calculus is refutationally complete and is strong enough to capture some standard restrictions of the superposition calculus. We discuss the implementation of the calculus in the theorem prover Vampire. Our experiments show the power of the new approach: we were able to solve 24 TPTP problems not previously solved by any prover, including previous versions of Vampire. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:40:02 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hajdu",
"Márton",
""
],
[
"Kovács",
"Laura",
""
],
[
"Voronkov",
"Andrei",
""
]
] |
2505.22214 | Michal Sojka | Ond\v{r}ej Benedikt, Michal Sojka, P\v{r}emysl \v{S}\r{u}cha, Pavel Zaykov, Zden\v{e}k Hanz\'alek | Thermal Modeling and Optimal Allocation of Avionics Safety-critical Tasks on Heterogeneous MPSoCs | null | null | 10.1016/j.jpdc.2025.105107 | null | cs.SE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Multi-Processor Systems-on-Chip (MPSoC) can deliver high performance needed in many industrial domains, including aerospace. However, their high power consumption, combined with avionics safety standards, brings new thermal management challenges. This paper investigates techniques for offline thermal-aware allocation of periodic tasks on heterogeneous MPSoCs running at a fixed clock frequency, as required in avionics. The goal is to find the assignment of tasks to (i) cores and (ii) temporal isolation windows while minimizing the MPSoC temperature. To achieve that, we propose and analyze three power models, and integrate them within several novel optimization approaches based on heuristics, a black-box optimizer, and Integer Linear Programming (ILP). We perform the experimental evaluation on three popular MPSoC platforms (NXP i.MX8QM MEK, NXP i.MX8QM Ixora, NVIDIA TX2) and observe a difference of up to 5.5{\deg}C among the tested methods (corresponding to a 22% reduction w.r.t. the ambient temperature). We also show that our method, integrating the empirical power model with the ILP, outperforms the other methods on all tested platforms. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:40:47 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Benedikt",
"Ondřej",
""
],
[
"Sojka",
"Michal",
""
],
[
"Šůcha",
"Přemysl",
""
],
[
"Zaykov",
"Pavel",
""
],
[
"Hanzálek",
"Zdeněk",
""
]
] |
2505.22215 | Yihan He | Yihan He, Ming-Chun Hong, Wanli Zheng, Ching Shih, Hsin-Han Lee, Yu-Chen Hsin, Jeng-Hua Wei, Xiao Gong, Tuo-Hung Hou, and Gengchiau Liang | Solving Boolean Satisfiability Problems Using A Hypergraph-based Probabilistic Computer | 17 pages, 7 figures | null | null | null | physics.comp-ph | http://creativecommons.org/licenses/by/4.0/ | Boolean Satisfiability (SAT) problems are critical in fields such as artificial intelligence and cryptography, where efficient solutions are essential. Conventional probabilistic solvers often encounter scalability issues due to complex logic synthesis steps. In this work, we present a novel approach for solving the 3-SAT Boolean satisfiability problem using hypergraph-based probabilistic computers obtained through direct mapping. This method directly translates 3-SAT logical expressions into hypergraph structures, thereby circumventing conventional logic decomposition and synthesis procedures, and offering a more streamlined solver architecture. For a uf20-01 instance, our approach significantly reduces the vertex number from 112 to 20 with a reduced solution space from 2112 to 220. Numerical simulations demonstrate that the proposed hypergraph-based solver achieves a significantly higher success rate of up to 99%, compared to merely 1% for conventional solvers. Furthermore, the proposed direct mapping method can be extended to solve k-SAT problems, which provides a scalable framework for tackling more complex satisfiability problems using probabilistic computing in the future. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:48:19 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"He",
"Yihan",
""
],
[
"Hong",
"Ming-Chun",
""
],
[
"Zheng",
"Wanli",
""
],
[
"Shih",
"Ching",
""
],
[
"Lee",
"Hsin-Han",
""
],
[
"Hsin",
"Yu-Chen",
""
],
[
"Wei",
"Jeng-Hua",
""
],
[
"Gong",
"Xiao",
""
],
[
"Hou",
"Tuo-Hung",
""
],
[
"Liang",
"Gengchiau",
""
]
] |
2505.22216 | Lucas Hallgren Mr | Lucas Hallgren, Radoslaw Wojtak, Jens Hjorth, Charles L. Steinhardt | Diverse reddening distributions in sight lines to type Ia supernovae | Submitted to A&A; 12 pages, 6 figures, 3 tables | null | null | null | astro-ph.GA astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | Precise cosmological constraints from type Ia supernovae require adequately accurate corrections for host-galaxy extinction. Modelling these corrections is challenged by the problem of disentangling supernova intrinsic colours from host-galaxy interstellar reddening.The latter is commonly modelled in a probabilistic way assuming an exponential distribution exp(-E(B-V)/\tau) as a universal prior which is applied across all types of supernova host galaxies. We test the robustness of the exponential model and its universality against predictions based on simulating dust and type Ia supernova distributions in host galaxies of different morphological types. We find substantial differences between predicted interstellar reddening in late- and early-type host galaxies, primarily driven by the stellar-to-dust mass ratios. The mean simulated reddening in late-type galaxies matches well those derived from type Ia supernova observations, but it is significantly lower for early-type host galaxies. The reddening distributions exhibit an excess of sight lines with vanishing reddening with respect to the exponential model, although the difference is quite mild for late-type galaxies. On the other hand, the distribution may peak at E(B-V)>0 when considering a population of young type Ia supernovae originating from lower heights within the dust disc. We demonstrate that assuming a universal reddening prior distribution for modeling peak magnitude-colour relation, which is currently a common practice, gives rise to a spurious scatter in the derived extinction properties. It may also bias relative distances between supernovae originating from different host-galaxy populations. The discrepancy between the simulated reddening in average early-type host galaxies and the observed occurrence of reddened supernovae suggests that reddening does not originate from interstellar dust expected in these galaxies. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:49:36 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hallgren",
"Lucas",
""
],
[
"Wojtak",
"Radoslaw",
""
],
[
"Hjorth",
"Jens",
""
],
[
"Steinhardt",
"Charles L.",
""
]
] |
2505.22217 | Sean Adamson | Sean A. Adamson and Petros Wallden | Benincasa-Dowker causal set actions by quantum counting | 22 pages, 5 figures | null | null | null | quant-ph gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Causal set theory is an approach to quantum gravity in which spacetime is fundamentally discrete while retaining local Lorentz invariance. The Benincasa--Dowker action is the causal set equivalent to the Einstein--Hilbert action underpinning Einstein's general theory of relativity. We present a $\tilde{O}(n^{2})$ running-time quantum algorithm to compute the Benincasa--Dowker action in arbitrary spacetime dimensions for causal sets with $n$ elements which is asymptotically optimal and offers a polynomial speed up compared to all known classical or quantum algorithms. To do this, we prepare a uniform superposition over an $O(n^{2})$-size arbitrary subset of computational basis states encoding the classical description of a causal set of interest. We then construct $\tilde{O}(n)$-depth oracle circuits testing for different discrete volumes between pairs of causal set elements. Repeatedly performing a two-stage variant of quantum counting using these oracles yields the desired algorithm. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:50:06 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Adamson",
"Sean A.",
""
],
[
"Wallden",
"Petros",
""
]
] |
2505.22218 | Ji\v{r}\'i Ajgl | Ji\v{r}\'i Ajgl, Ond\v{r}ej Straka | Aspects of density approximation by tensor trains | Submitted to the conference FUSION 2025 | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | Point-mass filters solve Bayesian recursive relations by approximating probability density functions of a system state over grids of discrete points. The approach suffers from the curse of dimensionality. The exponential increase of the number of the grid points can be mitigated by application of low-rank approximations of multidimensional arrays. Tensor train decompositions represent individual values by the product of matrices. This paper focuses on selected issues that are substantial in state estimation. Namely, the contamination of the density approximations by negative values is discussed first. Functional decompositions of quadratic functions are compared with decompositions of discretised Gaussian densities next. In particular, the connection of correlation with tensor train ranks is explored. Last, the consequences of interpolating the density values from one grid to a new grid are analysed. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:50:50 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ajgl",
"Jiří",
""
],
[
"Straka",
"Ondřej",
""
]
] |
2505.22219 | Subhadeep Mukherjee | Subhadeep Mukherjee, Shashank Shekhar Pandey, A. S. Majumdar | Constraining the Hubble parameter with the 21 cm brightness temperature signal in a universe with inhomogeneities | 14 pages , 3 figures | null | null | null | astro-ph.CO gr-qc | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We consider the 21\,cm brightness temperature as a probe of the Hubble tension in the framework of an inhomogeneous cosmological model. Employing Buchert's averaging formalism to study the effect of inhomogeneities on the background evolution, we consider scaling laws for the backreaction and curvature consistent with structure formation simulations. We calibrate the effective matter density using MCMC analysis using Union 2.1 Supernova Ia data. Our results show that a higher Hubble constant ($\sim73$\,km/s/Mpc) leads to a shallower absorption feature in the brightness temperature versus redshift curve. On the other hand, a lower value ($\sim67$\,km/s/Mpc) produces a remarkable dip in the brightness temperature $T_{21}$. Such a substantial difference is absent in the standard $\Lambda$CDM model. Our findings indicate that inhomogeneities could significantly affect the 21\,cm signal, and may shed further light on the different measurements of the Hubble constant. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:50:55 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mukherjee",
"Subhadeep",
""
],
[
"Pandey",
"Shashank Shekhar",
""
],
[
"Majumdar",
"A. S.",
""
]
] |
2505.22220 | Sebastian Zillien | Denis Petrov, Pascal Ruffing, Sebastian Zillien, Steffen Wendzel | Domainator: Detecting and Identifying DNS-Tunneling Malware Using Metadata Sequences | null | null | null | null | cs.CR cs.NI | http://creativecommons.org/licenses/by/4.0/ | In recent years, malware with tunneling (or: covert channel) capabilities is on the rise. While malware research led to several methods and innovations, the detection and differentiation of malware solely based on its DNS tunneling features is still in its infancy. Moreover, no work so far has used the DNS tunneling traffic to gain knowledge over the current actions taken by the malware. In this paper, we present Domainator, an approach to detect and differentiate state-of-the-art malware and DNS tunneling tools without relying on trivial (but quickly altered) features such as "magic bytes" that are embedded into subdomains. Instead, we apply an analysis of sequential patterns to identify specific types of malware. We evaluate our approach with 7 different malware samples and tunneling tools and can identify the particular malware based on its DNS traffic. We further infer the rough behavior of the particular malware through its DNS tunneling artifacts. Finally, we compare our Domainator with related methods. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:52:19 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Petrov",
"Denis",
""
],
[
"Ruffing",
"Pascal",
""
],
[
"Zillien",
"Sebastian",
""
],
[
"Wendzel",
"Steffen",
""
]
] |
2505.22221 | Stephan Sponar | Andreas Dvorak, Ismaele V. Masiello, Yuji Hasegawa, Hartmut Lemmel, Holger F. Hofmann, Stephan Sponar | Tight qubit uncertainty relations studied through weak values in neutron interferometry | 10 pages, 8 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | In its original formulation, Heisenberg's uncertainty principle describes a trade-off relation between the error of a quantum measurement and the thereby induced disturbance on the measured object. However, this relation is not valid in general. An alternative universally valid relation was derived by Ozawa in 2003, defining error and disturbance in a general concept, experimentally accessible via a tomographic method. Later, it was shown by Hall that these errors correspond to the statistical deviation between a physical property and its estimate. Recently, it was discovered that these errors can be observed experimentally when weak values are determined through a procedure named "feedback compensation". Here, we apply this procedure for the complete experimental characterization of the error-disturbance relation between a which-way observable in an interferometer and another observable associated with the output of the interferometer, confirming the theoretically predicted relation. As expected for pure states, the uncertainty is tightly fulfilled. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:54:31 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Dvorak",
"Andreas",
""
],
[
"Masiello",
"Ismaele V.",
""
],
[
"Hasegawa",
"Yuji",
""
],
[
"Lemmel",
"Hartmut",
""
],
[
"Hofmann",
"Holger F.",
""
],
[
"Sponar",
"Stephan",
""
]
] |
2505.22222 | Yunsoo Kim | Yunsoo Kim, Jinge Wu, Su-Hwan Kim, Pardeep Vasudev, Jiashu Shen, Honghan Wu | Look & Mark: Leveraging Radiologist Eye Fixations and Bounding boxes in Multimodal Large Language Models for Chest X-ray Report Generation | null | null | null | null | cs.CV cs.CL | http://creativecommons.org/licenses/by/4.0/ | Recent advancements in multimodal Large Language Models (LLMs) have significantly enhanced the automation of medical image analysis, particularly in generating radiology reports from chest X-rays (CXR). However, these models still suffer from hallucinations and clinically significant errors, limiting their reliability in real-world applications. In this study, we propose Look & Mark (L&M), a novel grounding fixation strategy that integrates radiologist eye fixations (Look) and bounding box annotations (Mark) into the LLM prompting framework. Unlike conventional fine-tuning, L&M leverages in-context learning to achieve substantial performance gains without retraining. When evaluated across multiple domain-specific and general-purpose models, L&M demonstrates significant gains, including a 1.2% improvement in overall metrics (A.AVG) for CXR-LLaVA compared to baseline prompting and a remarkable 9.2% boost for LLaVA-Med. General-purpose models also benefit from L&M combined with in-context learning, with LLaVA-OV achieving an 87.3% clinical average performance (C.AVG)-the highest among all models, even surpassing those explicitly trained for CXR report generation. Expert evaluations further confirm that L&M reduces clinically significant errors (by 0.43 average errors per report), such as false predictions and omissions, enhancing both accuracy and reliability. These findings highlight L&M's potential as a scalable and efficient solution for AI-assisted radiology, paving the way for improved diagnostic workflows in low-resource clinical settings. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:54:40 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kim",
"Yunsoo",
""
],
[
"Wu",
"Jinge",
""
],
[
"Kim",
"Su-Hwan",
""
],
[
"Vasudev",
"Pardeep",
""
],
[
"Shen",
"Jiashu",
""
],
[
"Wu",
"Honghan",
""
]
] |
2505.22223 | Linfeng Mei | Wang-Ji Yan, Lin-Feng Mei, Yuan-Wei Yin, Jiang Mo, Costas Papadimitriou, Ka-Veng Yuen, Michael Beer | Bayesian Learning in Structural Dynamics: A Comprehensive Review and Emerging Trends | 151 pages, 9 figures | null | null | null | physics.data-an physics.comp-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Bayesian learning has emerged as a compelling and vital research direction in the field of structural dynamics, offering a probabilistic lens to understand and refine the analysis of complex dynamical systems. This review meticulously traces the three-decade evolution of Bayesian learning in structural dynamics, illuminating core principles, groundbreaking methodologies, and diverse applications that have significantly influenced the field. The narrative commences by delving into the basics of Bayesian theory, clarifying essential concepts, and introducing primary methods for deriving posterior distributions, with an in-depth exploration of three types: Laplace approximation, stochastic sampling, and variational inference. Subsequently, the text explores the implementation of two types of Bayesian learning in structural dynamics: physical model learning and data-centric statistical model learning. Physical model learning emphasizes inferring physical model parameters within a Bayesian framework for system identification and prediction, while statistical model learning integrates Bayesian learning methodologies into data-centric statistical modeling within probabilistic machine learning. Both types resonate across various applications, such as modal analysis, model updating, damage detection, and reliability updating, highlighting their pivotal role in enhancing comprehension of dynamical systems and decision-making. The paper also navigates obstacles by proposing ways to enhance existing Bayesian inference strategies. Distinguished from previous research, this study offers a thorough examination of both traditional and cutting-edge Bayesian methods. It not only underscores the transformative influence of Bayesian approaches but also serves as a beacon, guiding researchers in the judicious selection and refinement of suitable methods for various challenges in structural dynamics. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:54:41 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Yan",
"Wang-Ji",
""
],
[
"Mei",
"Lin-Feng",
""
],
[
"Yin",
"Yuan-Wei",
""
],
[
"Mo",
"Jiang",
""
],
[
"Papadimitriou",
"Costas",
""
],
[
"Yuen",
"Ka-Veng",
""
],
[
"Beer",
"Michael",
""
]
] |
2505.22224 | Senne Berden | Senne Berden, Ali \.Irfan Mahmuto\u{g}ullar{\i}, Dimos Tsouros, Tias Guns | Solver-Free Decision-Focused Learning for Linear Optimization Problems | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Mathematical optimization is a fundamental tool for decision-making in a wide range of applications. However, in many real-world scenarios, the parameters of the optimization problem are not known a priori and must be predicted from contextual features. This gives rise to predict-then-optimize problems, where a machine learning model predicts problem parameters that are then used to make decisions via optimization. A growing body of work on decision-focused learning (DFL) addresses this setting by training models specifically to produce predictions that maximize downstream decision quality, rather than accuracy. While effective, DFL is computationally expensive, because it requires solving the optimization problem with the predicted parameters at each loss evaluation. In this work, we address this computational bottleneck for linear optimization problems, a common class of problems in both DFL literature and real-world applications. We propose a solver-free training method that exploits the geometric structure of linear optimization to enable efficient training with minimal degradation in solution quality. Our method is based on the insight that a solution is optimal if and only if it achieves an objective value that is at least as good as that of its adjacent vertices on the feasible polytope. Building on this, our method compares the estimated quality of the ground-truth optimal solution with that of its precomputed adjacent vertices, and uses this as loss function. Experiments demonstrate that our method significantly reduces computational cost while maintaining high decision quality. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:55:16 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Berden",
"Senne",
""
],
[
"Mahmutoğulları",
"Ali İrfan",
""
],
[
"Tsouros",
"Dimos",
""
],
[
"Guns",
"Tias",
""
]
] |
2505.22225 | Yuan Feng | Yuan Feng, Ye-Fei Yuan, Shuang-Nan Zhang | Reflection Spectra of Accretion Disks Illuminated by an Off-Axis Corona | 15 pages, 12 figures | Astrophysical Journal 984, 173 (2025) | 10.3847/1538-4357/adc8a6 | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Relativistic reflection features in the X-ray spectra of accreting black holes are considered to be generated by the illumination of the accretion disk by the hot corona. In this work, we present a numerical method for the emission line profile and the reflection spectrum produced by an off-axis X-ray source. The X-ray source is considered as a point source, as in the lamppost scenario, except that it is located off-axis and moves at arbitrary velocity. The observed flux for the distant observer is calculated directly without priority evaluation of the emissivity on the accretion disk, which allows our model to be applicable to the point source that deviates from the axis of the black hole spins and moves with a velocity. To study the impact of the off-axis geometry on the measurement of source properties, we simulate observations for a black hole binary with NuSTAR and eXTP. We compare the simulation with the observation of the phase-resolved spectra of the low-frequency quasiperiodic oscillation observed by the Insight Hard X-ray Modulation Telescope. Due to the nonaxisymmetric illumination on the accretion disk, parameters of the model are not reproduced by the lamppost model, including the corona height, radial velocity, and the reflection fraction. On the other hand, all the model parameters are recovered through the off-axis model. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:58:43 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Feng",
"Yuan",
""
],
[
"Yuan",
"Ye-Fei",
""
],
[
"Zhang",
"Shuang-Nan",
""
]
] |
2505.22226 | Xuyang Zhang | Xuyang Zhang, Xi Zhang, Liang Chen, Hao Shi, Qingshan Guo | Hadaptive-Net: Efficient Vision Models via Adaptive Cross-Hadamard Synergy | 10 pages, 5 figures | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent studies have revealed the immense potential of Hadamard product in enhancing network representational capacity and dimensional compression. However, despite its theoretical promise, this technique has not been systematically explored or effectively applied in practice, leaving its full capabilities underdeveloped. In this work, we first analyze and identify the advantages of Hadamard product over standard convolutional operations in cross-channel interaction and channel expansion. Building upon these insights, we propose a computationally efficient module: Adaptive Cross-Hadamard (ACH), which leverages adaptive cross-channel Hadamard products for high-dimensional channel expansion. Furthermore, we introduce Hadaptive-Net (Hadamard Adaptive Network), a lightweight network backbone for visual tasks, which is demonstrated through experiments that it achieves an unprecedented balance between inference speed and accuracy through our proposed module. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 10:58:56 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Xuyang",
""
],
[
"Zhang",
"Xi",
""
],
[
"Chen",
"Liang",
""
],
[
"Shi",
"Hao",
""
],
[
"Guo",
"Qingshan",
""
]
] |
2505.22227 | Carlos Mera Acosta | Carlos Mera Acosta | Unified Magnetoelectric Mechanism for Spin Splitting in Magnets | 6 pages and 1 figure | null | null | null | cond-mat.str-el cond-mat.other | http://creativecommons.org/licenses/by/4.0/ | We identify a relativistic magnetoelectric correction that unifies the understanding of spin splitting (SS) in ferromagnets, antiferromagnets, and altermagnets. Based on Dirac theory, we show that local electric multipoles coupled to magnetic moments govern SS across compensated magnets, leading to quadrupole-, dipole-, or monopole-driven regimes. Our model predicts momentum-dependent and momentum-independent SS, reconciling seemingly distinct phenomena such as altermagnetism and the spin Zeeman effect. This work establishes a general physical mechanism for SS beyond conventional Zeeman and Rashba paradigms. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:02:37 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Acosta",
"Carlos Mera",
""
]
] |
2505.22228 | Haibin He | Haibin He, Jing Zhang, Maoyuan Ye, Juhua Liu, Bo Du, Dacheng Tao | GoMatching++: Parameter- and Data-Efficient Arbitrary-Shaped Video Text Spotting and Benchmarking | arXiv admin note: text overlap with arXiv:2401.07080 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Video text spotting (VTS) extends image text spotting (ITS) by adding text tracking, significantly increasing task complexity. Despite progress in VTS, existing methods still fall short of the performance seen in ITS. This paper identifies a key limitation in current video text spotters: limited recognition capability, even after extensive end-to-end training. To address this, we propose GoMatching++, a parameter- and data-efficient method that transforms an off-the-shelf image text spotter into a video specialist. The core idea lies in freezing the image text spotter and introducing a lightweight, trainable tracker, which can be optimized efficiently with minimal training data. Our approach includes two key components: (1) a rescoring mechanism to bridge the domain gap between image and video data, and (2) the LST-Matcher, which enhances the frozen image text spotter's ability to handle video text. We explore various architectures for LST-Matcher to ensure efficiency in both parameters and training data. As a result, GoMatching++ sets new performance records on challenging benchmarks such as ICDAR15-video, DSText, and BOVText, while significantly reducing training costs. To address the lack of curved text datasets in VTS, we introduce ArTVideo, a new benchmark featuring over 30% curved text with detailed annotations. We also provide a comprehensive statistical analysis and experimental results for ArTVideo. We believe that GoMatching++ and the ArTVideo benchmark will drive future advancements in video text spotting. The source code, models and dataset are publicly available at https://github.com/Hxyz-123/GoMatching. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:02:45 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"He",
"Haibin",
""
],
[
"Zhang",
"Jing",
""
],
[
"Ye",
"Maoyuan",
""
],
[
"Liu",
"Juhua",
""
],
[
"Du",
"Bo",
""
],
[
"Tao",
"Dacheng",
""
]
] |
2505.22229 | Zixuan Li | Zixuan Li and Xueliang Zhang and Lei Miao and Zhipeng Yan | Two-stage Audio-Visual Target Speaker Extraction System for Real-Time Processing On Edge Device | null | null | null | null | cs.SD eess.AS | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Audio-Visual Target Speaker Extraction (AVTSE) aims to isolate a target speaker's voice in a multi-speaker environment with visual cues as auxiliary. Most of the existing AVTSE methods encode visual and audio features simultaneously, resulting in extremely high computational complexity and making it impractical for real-time processing on edge devices. To tackle this issue, we proposed a two-stage ultra-compact AVTSE system. Specifically, in the first stage, a compact network is employed for voice activity detection (VAD) using visual information. In the second stage, the VAD results are combined with audio inputs to isolate the target speaker's voice. Experiments show that the proposed system effectively suppresses background noise and interfering voices while spending little computational resources. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:05:24 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Li",
"Zixuan",
""
],
[
"Zhang",
"Xueliang",
""
],
[
"Miao",
"Lei",
""
],
[
"Yan",
"Zhipeng",
""
]
] |
2505.22230 | Zhisong Wang | Zhisong Wang, Yiwen Ye, Ziyang Chen, and Yong Xia | Enjoying Information Dividend: Gaze Track-based Medical Weakly Supervised Segmentation | 10 pages, 4 figures, MICCAI 2025 (Early Accept) | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Weakly supervised semantic segmentation (WSSS) in medical imaging struggles with effectively using sparse annotations. One promising direction for WSSS leverages gaze annotations, captured via eye trackers that record regions of interest during diagnostic procedures. However, existing gaze-based methods, such as GazeMedSeg, do not fully exploit the rich information embedded in gaze data. In this paper, we propose GradTrack, a framework that utilizes physicians' gaze track, including fixation points, durations, and temporal order, to enhance WSSS performance. GradTrack comprises two key components: Gaze Track Map Generation and Track Attention, which collaboratively enable progressive feature refinement through multi-level gaze supervision during the decoding process. Experiments on the Kvasir-SEG and NCI-ISBI datasets demonstrate that GradTrack consistently outperforms existing gaze-based methods, achieving Dice score improvements of 3.21\% and 2.61\%, respectively. Moreover, GradTrack significantly narrows the performance gap with fully supervised models such as nnUNet. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:05:50 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Wang",
"Zhisong",
""
],
[
"Ye",
"Yiwen",
""
],
[
"Chen",
"Ziyang",
""
],
[
"Xia",
"Yong",
""
]
] |
2505.22231 | Stefan Bleeck | Stefan Bleeck | Advancing Hearing Assessment: An ASR-Based Frequency-Specific Speech Test for Diagnosing Presbycusis | null | null | null | null | cs.SD cs.CL eess.AS | http://creativecommons.org/licenses/by/4.0/ | Traditional audiometry often fails to fully characterize the functional impact of hearing loss on speech understanding, particularly supra-threshold deficits and frequency-specific perception challenges in conditions like presbycusis. This paper presents the development and simulated evaluation of a novel Automatic Speech Recognition (ASR)-based frequency-specific speech test designed to provide granular diagnostic insights. Our approach leverages ASR to simulate the perceptual effects of moderate sloping hearing loss by processing speech stimuli under controlled acoustic degradation and subsequently analyzing phoneme-level confusion patterns. Key findings indicate that simulated hearing loss introduces specific phoneme confusions, predominantly affecting high-frequency consonants (e.g., alveolar/palatal to labiodental substitutions) and leading to significant phoneme deletions, consistent with the acoustic cues degraded in presbycusis. A test battery curated from these ASR-derived confusions demonstrated diagnostic value, effectively differentiating between simulated normal-hearing and hearing-impaired listeners in a comprehensive simulation. This ASR-driven methodology offers a promising avenue for developing objective, granular, and frequency-specific hearing assessment tools that complement traditional audiometry. Future work will focus on validating these findings with human participants and exploring the integration of advanced AI models for enhanced diagnostic precision. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:06:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bleeck",
"Stefan",
""
]
] |
2505.22232 | Manuel Brack | Mehdi Ali, Manuel Brack, Max L\"ubbering, Elias Wendt, Abbas Goher Khan, Richard Rutmann, Alex Jude, Maurice Kraus, Alexander Arno Weber, Felix Stollenwerk, David Kacz\'er, Florian Mai, Lucie Flek, Rafet Sifa, Nicolas Flores-Herr, Joachim K\"ohler, Patrick Schramowski, Michael Fromm, Kristian Kersting | Judging Quality Across Languages: A Multilingual Approach to Pretraining Data Filtering with Language Models | Project page available at https://huggingface.co/spaces/Jackal-AI/JQL | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | High-quality multilingual training data is essential for effectively pretraining large language models (LLMs). Yet, the availability of suitable open-source multilingual datasets remains limited. Existing state-of-the-art datasets mostly rely on heuristic filtering methods, restricting both their cross-lingual transferability and scalability. Here, we introduce JQL, a systematic approach that efficiently curates diverse and high-quality multilingual data at scale while significantly reducing computational demands. JQL distills LLMs' annotation capabilities into lightweight annotators based on pretrained multilingual embeddings. These models exhibit robust multilingual and cross-lingual performance, even for languages and scripts unseen during training. Evaluated empirically across 35 languages, the resulting annotation pipeline substantially outperforms current heuristic filtering methods like Fineweb2. JQL notably enhances downstream model training quality and increases data retention rates. Our research provides practical insights and valuable resources for multilingual data curation, raising the standards of multilingual dataset development. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:06:54 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ali",
"Mehdi",
""
],
[
"Brack",
"Manuel",
""
],
[
"Lübbering",
"Max",
""
],
[
"Wendt",
"Elias",
""
],
[
"Khan",
"Abbas Goher",
""
],
[
"Rutmann",
"Richard",
""
],
[
"Jude",
"Alex",
""
],
[
"Kraus",
"Maurice",
""
],
[
"Weber",
"Alexander Arno",
""
],
[
"Stollenwerk",
"Felix",
""
],
[
"Kaczér",
"David",
""
],
[
"Mai",
"Florian",
""
],
[
"Flek",
"Lucie",
""
],
[
"Sifa",
"Rafet",
""
],
[
"Flores-Herr",
"Nicolas",
""
],
[
"Köhler",
"Joachim",
""
],
[
"Schramowski",
"Patrick",
""
],
[
"Fromm",
"Michael",
""
],
[
"Kersting",
"Kristian",
""
]
] |
2505.22233 | Ugo Bruzzo | Ugo Bruzzo and Daniel Hern\'andez Ruip\'erez | Moduli of stable supermaps | 21 pages | null | null | null | math.AG hep-th math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We review the notion of stable supermap from SUSY curves to a fixed target superscheme, and prove that when the target is (super)projective, stable supermaps are parameterized by an algebraic superstack with superschematic and separated diagonal. We characterize the bosonic reduction of this moduli superstack and see that it has a surjective morphism onto the moduli stack of stable maps from spin curves to the bosonic reduction of the target, whose fibers are linear schemes; for this reason, the moduli superstack of stable supermaps is not proper unless such linear schemes reduce to a point. Using Manin-Penkov-Voronov's super Grothendieck-Riemann-Roch theorem we also make a formal computation of the virtual dimension of the moduli superstack, which agrees with the characterization of the bosonic reduction just mentioned and with the dimension formula for the case of bosonic target existing in the literature. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:08:49 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bruzzo",
"Ugo",
""
],
[
"Ruipérez",
"Daniel Hernández",
""
]
] |
2505.22234 | Georgia Kapitsaki | Georgia M. Kapitsaki, Maria Papoutsoglou | Evolution of repositories and privacy laws: commit activities in the GDPR and CCPA era | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Free and open source software has gained a lot of momentum in the industry and the research community. The latest advances in privacy legislation, including the EU General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA), have forced the community to pay special attention to users' data privacy. The main aim of this work is to examine software repositories that are acting on privacy laws. We have collected commit data from GitHub repositories in order to understand indications on main data privacy laws (GDPR, CCPA, CPRA, UK DPA) in the last years. Via an automated process, we analyzed 37,213 commits from 12,391 repositories since 2016, whereas 594 commits from the 70 most popular repositories of the dataset were manually analyzed. We observe that most commits were performed on the year the law came into effect and privacy relevant terms appear in the commit messages, whereas reference to specific data privacy user rights is scarce. The study showed that more educational activities on data privacy user rights are needed, as well as tools for privacy recommendations, whereas verifying actual compliance via source code execution is a useful direction for software engineering researchers. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:10:58 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kapitsaki",
"Georgia M.",
""
],
[
"Papoutsoglou",
"Maria",
""
]
] |
2505.22235 | Amon Lahr | Amon Lahr, Johannes K\"ohler, Anna Scampicchio, Melanie N. Zeilinger | Optimal kernel regression bounds under energy-bounded noise | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Non-conservative uncertainty bounds are key for both assessing an estimation algorithm's accuracy and in view of downstream tasks, such as its deployment in safety-critical contexts. In this paper, we derive a tight, non-asymptotic uncertainty bound for kernel-based estimation, which can also handle correlated noise sequences. Its computation relies on a mild norm-boundedness assumption on the unknown function and the noise, returning the worst-case function realization within the hypothesis class at an arbitrary query input location. The value of this function is shown to be given in terms of the posterior mean and covariance of a Gaussian process for an optimal choice of the measurement noise covariance. By rigorously analyzing the proposed approach and comparing it with other results in the literature, we show its effectiveness in returning tight and easy-to-compute bounds for kernel-based estimates. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:11:24 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lahr",
"Amon",
""
],
[
"Köhler",
"Johannes",
""
],
[
"Scampicchio",
"Anna",
""
],
[
"Zeilinger",
"Melanie N.",
""
]
] |
2505.22236 | Charlotte Pouw | Charlotte Pouw, Afra Alishahi, Willem Zuidema | A Linguistically Motivated Analysis of Intonational Phrasing in Text-to-Speech Systems: Revealing Gaps in Syntactic Sensitivity | Accepted to CoNLL 2025 | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | We analyze the syntactic sensitivity of Text-to-Speech (TTS) systems using methods inspired by psycholinguistic research. Specifically, we focus on the generation of intonational phrase boundaries, which can often be predicted by identifying syntactic boundaries within a sentence. We find that TTS systems struggle to accurately generate intonational phrase boundaries in sentences where syntactic boundaries are ambiguous (e.g., garden path sentences or sentences with attachment ambiguity). In these cases, systems need superficial cues such as commas to place boundaries at the correct positions. In contrast, for sentences with simpler syntactic structures, we find that systems do incorporate syntactic cues beyond surface markers. Finally, we finetune models on sentences without commas at the syntactic boundary positions, encouraging them to focus on more subtle linguistic cues. Our findings indicate that this leads to more distinct intonation patterns that better reflect the underlying structure. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:11:29 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Pouw",
"Charlotte",
""
],
[
"Alishahi",
"Afra",
""
],
[
"Zuidema",
"Willem",
""
]
] |
2505.22237 | Fatma Kader Bing\"ol | Fatma Kader Bing\"ol, Adam Chapman and Ahmed Laghribi | Essential dimension of sequences of quaternion algebras | null | null | null | null | math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the essential dimension of the set of isometry classes of $m$-tuples $(\varphi_1,...,\varphi_m)$ of quadratic $n$-fold Pfister forms over a field $F$ such that the Witt class of $\varphi_1 \perp \ldots \perp \varphi_m$ lies in $I_q^{n+1}F$. We show that the essential dimension is equal to $n+1$, when $m=3$, and is either $4$ or $5$, when $n=\text{char} F=2$, $m=4$. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:12:23 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bingöl",
"Fatma Kader",
""
],
[
"Chapman",
"Adam",
""
],
[
"Laghribi",
"Ahmed",
""
]
] |
2505.22238 | Alexander Ploshkin | A. Ploshkin, V. Tytskiy, A. Pismenny, V. Baikalov, E. Taychinov, A. Permiakov, D. Burlakov, E. Krofto, N. Savushkin | Yambda-5B -- A Large-Scale Multi-modal Dataset for Ranking And Retrieval | null | null | null | null | cs.IR cs.LG | http://creativecommons.org/licenses/by/4.0/ | We present Yambda-5B, a large-scale open dataset sourced from the Yandex.Music streaming platform. Yambda-5B contains 4.79 billion user-item interactions from 1 million users across 9.39 million tracks. The dataset includes two primary types of interactions: implicit feedback (listening events) and explicit feedback (likes, dislikes, unlikes and undislikes). In addition, we provide audio embeddings for most tracks, generated by a convolutional neural network trained on audio spectrograms. A key distinguishing feature of Yambda-5B is the inclusion of the is_organic flag, which separates organic user actions from recommendation-driven events. This distinction is critical for developing and evaluating machine learning algorithms, as Yandex.Music relies on recommender systems to personalize track selection for users. To support rigorous benchmarking, we introduce an evaluation protocol based on a Global Temporal Split, allowing recommendation algorithms to be assessed in conditions that closely mirror real-world use. We report benchmark results for standard baselines (ItemKNN, iALS) and advanced models (SANSA, SASRec) using a variety of evaluation metrics. By releasing Yambda-5B to the community, we aim to provide a readily accessible, industrial-scale resource to advance research, foster innovation, and promote reproducible results in recommender systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:12:57 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ploshkin",
"A.",
""
],
[
"Tytskiy",
"V.",
""
],
[
"Pismenny",
"A.",
""
],
[
"Baikalov",
"V.",
""
],
[
"Taychinov",
"E.",
""
],
[
"Permiakov",
"A.",
""
],
[
"Burlakov",
"D.",
""
],
[
"Krofto",
"E.",
""
],
[
"Savushkin",
"N.",
""
]
] |
2505.22239 | Darshan Raju | Darshan Raju, Mahinder Ramdin, Jean-Marc Simon, Peter Kruger, and Thijs J.H. Vlugt | Finite-size effects of the excess entropy computed from integrating the radial distribution function | null | null | null | null | cond-mat.stat-mech physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | Computation of the excess entropy from the second-order density expansion of the entropy holds strictly for infinite systems in the limit of small densities. For the reliable and efficient computation of excess entropy, it is important to understand finite-size effects. Here, expressions to compute excess entropy and Kirkwood-Buff (KB) integrals by integrating the Radial Distribution Function (RDF) in a finite volume are derived, from which Sex and KB integrals in the thermodynamic limit are obtained. The scaling of these integrals with system size is studied. We show that the integrals of excess entropy converge faster than KB integrals. We compute excess entropy from Monte Carlo simulations using the Wang-Ramirez-Dobnikar-Frenkel pair interaction potential by thermodynamic integration and by integration of the RDF. We show that excess entropy computed by integrating the RDF is identical to that of excess entropy computed from thermodynamic integration at low densities, provided the RDF is extrapolated to the thermodynamic limit. At higher densities, differences up to 20% are observed. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:17:02 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Raju",
"Darshan",
""
],
[
"Ramdin",
"Mahinder",
""
],
[
"Simon",
"Jean-Marc",
""
],
[
"Kruger",
"Peter",
""
],
[
"Vlugt",
"Thijs J. H.",
""
]
] |
2505.22240 | Yunsoo Kim | Yunsoo Kim, Yusuf Abdulle, Honghan Wu | BioHopR: A Benchmark for Multi-Hop, Multi-Answer Reasoning in Biomedical Domain | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Biomedical reasoning often requires traversing interconnected relationships across entities such as drugs, diseases, and proteins. Despite the increasing prominence of large language models (LLMs), existing benchmarks lack the ability to evaluate multi-hop reasoning in the biomedical domain, particularly for queries involving one-to-many and many-to-many relationships. This gap leaves the critical challenges of biomedical multi-hop reasoning underexplored. To address this, we introduce BioHopR, a novel benchmark designed to evaluate multi-hop, multi-answer reasoning in structured biomedical knowledge graphs. Built from the comprehensive PrimeKG, BioHopR includes 1-hop and 2-hop reasoning tasks that reflect real-world biomedical complexities.
Evaluations of state-of-the-art models reveal that O3-mini, a proprietary reasoning-focused model, achieves 37.93% precision on 1-hop tasks and 14.57% on 2-hop tasks, outperforming proprietary models such as GPT4O and open-source biomedical models including HuatuoGPT-o1-70B and Llama-3.3-70B. However, all models exhibit significant declines in multi-hop performance, underscoring the challenges of resolving implicit reasoning steps in the biomedical domain. By addressing the lack of benchmarks for multi-hop reasoning in biomedical domain, BioHopR sets a new standard for evaluating reasoning capabilities and highlights critical gaps between proprietary and open-source models while paving the way for future advancements in biomedical LLMs. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:19:01 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kim",
"Yunsoo",
""
],
[
"Abdulle",
"Yusuf",
""
],
[
"Wu",
"Honghan",
""
]
] |
2505.22241 | Xia Zhou | Xia Zhou, Mark Wallace, Daniel D. Harabor, and Zhenliang Ma | An Exact System Optimum Assignment Model for Transit Demand Management | 18 pages, 13 figures | null | null | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Mass transit systems are experiencing increasing congestion in many cities. The schedule-based transit assignment problem (STAP) involves a joint choice model for departure times and routes, defining a space-time path in which passengers decide when to depart and which route to take. User equilibrium (UE) models for the STAP indicates the current congestion cost, while a system optimum (SO) models can provide insights for congestion relief directions. However, current STAP methods rely on approximate SO (Approx. SO) models, which underestimate the potential for congestion reduction in the system. The few studies in STAP that compute exact SO solutions ignore realistic constraints such as hard capacity, multi-line networks, or spatial-temporal competing demand flows. The paper proposes an exact SO method for the STAP that overcomes these limitations. We apply our approach to a case study involving part of the Hong Kong Mass Transit Railway network, which includes 5 lines, 12 interacting origin-destination pairs and 52,717 passengers. Computing an Approx. SO solution for this system indicates a modest potential for congestion reduction measures, with a cost reduction of 17.39% from the UE solution. Our exact SO solution is 36.35% lower than the UE solution, which is more than double the potential for congestion reduction. We then show how the exact SO solution can be used to identify opportunities for congestion reduction: (i) which origin-destination pairs have the most potential to reduce congestion; (ii) how many passengers can be reasonably shifted; (iii) future system potential with increasing demand and expanding network capacity. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:22:41 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhou",
"Xia",
""
],
[
"Wallace",
"Mark",
""
],
[
"Harabor",
"Daniel D.",
""
],
[
"Ma",
"Zhenliang",
""
]
] |
2505.22242 | Jens Chluba | Jens Chluba, Geoffrey Vasil and Richard Battye | Ito calculus meets the Hubble tension: Effects of small-scale electron density fluctuations on the CMB anisotropies | 22 pages, 18 figures, submitted to MNRAS, comments welcome | null | null | null | astro-ph.CO astro-ph.GA hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we develop a novel formalism to include the effect of electron density fluctuations at ultra small scales (well below the sound horizon at last scattering) on the observed anisotropies of the Cosmic Microwave Background (CMB). We treat the electron field as an independent stochastic variable and obtain the required ensemble-averaged photon Boltzmann equations using Ito calculus. Beyond changes to the average recombination history (which can be incorporated in the standard approach) our work identifies two new effects caused by the clumpiness of the medium. The first is a correction to the Thomson visibility function caused by correlations of the electron fluctuations along the line of sight, leading to an additional broadening of the visibility towards higher redshifts which causes extra damping and smearing of the CMB anisotropies. The second effect is a reduction of the effective scattering rate in the (pre-)recombination era that affects the photon transfer functions in a non-trivial manner. These new effects are subdominant in LCDM but can be significant in cosmologies with an early onset of structure formation (e.g., due to generation of enhanced small-scale power) as suggested by a number of indicators (e.g., the abundance of high redshift galaxies observed by JWST). We discuss the relevance of these new effects to the Hubble tension, finding that corrections which cannot be captured by simple modifications to the average recombination history arise. This highlights how important an understanding of the recombination process is in cosmological inference, and that a coordinated simulation and analysis campaign is required as part of the search for the origin of the various tensions in cosmology. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:22:45 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Chluba",
"Jens",
""
],
[
"Vasil",
"Geoffrey",
""
],
[
"Battye",
"Richard",
""
]
] |
2505.22243 | Bin Li | Bin Li, Diwei Liu, Zehong Hu, Jia Jia | UDuo: Universal Dual Optimization Framework for Online Matching | null | null | null | null | cs.IR cs.LG | http://creativecommons.org/licenses/by/4.0/ | Online resource allocation under budget constraints critically depends on proper modeling of user arrival dynamics. Classical approaches employ stochastic user arrival models to derive near-optimal solutions through fractional matching formulations of exposed users for downstream allocation tasks. However, this is no longer a reasonable assumption when the environment changes dynamically. In this work, We propose the Universal Dual optimization framework UDuo, a novel paradigm that fundamentally rethinks online allocation through three key innovations: (i) a temporal user arrival representation vector that explicitly captures distribution shifts in user arrival patterns and resource consumption dynamics, (ii) a resource pacing learner with adaptive allocation policies that generalize to heterogeneous constraint scenarios, and (iii) an online time-series forecasting approach for future user arrival distributions that achieves asymptotically optimal solutions with constraint feasibility guarantees in dynamic environments. Experimental results show that UDuo achieves higher efficiency and faster convergence than the traditional stochastic arrival model in real-world pricing while maintaining rigorous theoretical validity for general online allocation problems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:25:50 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Li",
"Bin",
""
],
[
"Liu",
"Diwei",
""
],
[
"Hu",
"Zehong",
""
],
[
"Jia",
"Jia",
""
]
] |
2505.22244 | Yaron Halle | Yaron Halle, Ariel Felner, Sven Koenig and Oren Salzman | A Preprocessing Framework for Efficient Approximate Bi-Objective Shortest-Path Computation in the Presence of Correlated Objectives | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The bi-objective shortest-path (BOSP) problem seeks to find paths between start and target vertices of a graph while optimizing two conflicting objective functions. We consider the BOSP problem in the presence of correlated objectives. Such correlations often occur in real-world settings such as road networks, where optimizing two positively correlated objectives, such as travel time and fuel consumption, is common. BOSP is generally computationally challenging as the size of the search space is exponential in the number of objective functions and the graph size. Bounded sub-optimal BOSP solvers such as A*pex alleviate this complexity by approximating the Pareto-optimal solution set rather than computing it exactly (given a user-provided approximation factor). As the correlation between objective functions increases, smaller approximation factors are sufficient for collapsing the entire Pareto-optimal set into a single solution. We leverage this insight to propose an efficient algorithm that reduces the search effort in the presence of correlated objectives. Our approach for computing approximations of the entire Pareto-optimal set is inspired by graph-clustering algorithms. It uses a preprocessing phase to identify correlated clusters within a graph and to generate a new graph representation. This allows a natural generalization of A*pex to run up to five times faster on DIMACS dataset instances, a standard benchmark in the field. To the best of our knowledge, this is the first algorithm proposed that efficiently and effectively exploits correlations in the context of bi-objective search while providing theoretical guarantees on solution quality. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:26:14 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Halle",
"Yaron",
""
],
[
"Felner",
"Ariel",
""
],
[
"Koenig",
"Sven",
""
],
[
"Salzman",
"Oren",
""
]
] |
2505.22245 | Bangti Jin | Jiho Hong, Bangti Jin, Zhizhang Wu | Direct Algorithms for Reconstructing Small Conductivity Inclusions in Subdiffusion | 31 pages, 10 figures | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The subdiffusion model that involves a Caputo fractional derivative in time is widely used to describe anomalously slow diffusion processes. In this work we aim at recovering the locations of small conductivity inclusions in the model from boundary measurement, and develop novel direct algorithms based on the asymptotic expansion of the boundary measurement with respect to the size of the inclusions and approximate fundamental solutions. These algorithms involve only algebraic manipulations and are computationally cheap. To the best of our knowledge, they are first direct algorithms for the inverse conductivity problem in the context of the subdiffusion model. Moreover, we provide relevant theoretical underpinnings for the algorithms. Also we present numerical results to illustrate their performance under various scenarios, e.g., the size of inclusions, noise level of the data, and the number of inclusions, showing that the algorithms are efficient and robust. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:27:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hong",
"Jiho",
""
],
[
"Jin",
"Bangti",
""
],
[
"Wu",
"Zhizhang",
""
]
] |
2505.22246 | Nedko Savov | Nedko Savov, Naser Kazemi, Deheng Zhang, Danda Pani Paudel, Xi Wang, Luc Van Gool | StateSpaceDiffuser: Bringing Long Context to Diffusion World Models | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | World models have recently become promising tools for predicting realistic visuals based on actions in complex environments. However, their reliance on a short sequence of observations causes them to quickly lose track of context. As a result, visual consistency breaks down after just a few steps, and generated scenes no longer reflect information seen earlier. This limitation of the state-of-the-art diffusion-based world models comes from their lack of a lasting environment state. To address this problem, we introduce StateSpaceDiffuser, where a diffusion model is enabled to perform on long-context tasks by integrating a sequence representation from a state-space model (Mamba), representing the entire interaction history. This design restores long-term memory without sacrificing the high-fidelity synthesis of diffusion models. To rigorously measure temporal consistency, we develop an evaluation protocol that probes a model's ability to reinstantiate seen content in extended rollouts. Comprehensive experiments show that StateSpaceDiffuser significantly outperforms a strong diffusion-only baseline, maintaining a coherent visual context for an order of magnitude more steps. It delivers consistent views in both a 2D maze navigation and a complex 3D environment. These results establish that bringing state-space representations into diffusion models is highly effective in demonstrating both visual details and long-term memory. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:27:54 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Savov",
"Nedko",
""
],
[
"Kazemi",
"Naser",
""
],
[
"Zhang",
"Deheng",
""
],
[
"Paudel",
"Danda Pani",
""
],
[
"Wang",
"Xi",
""
],
[
"Van Gool",
"Luc",
""
]
] |
2505.22247 | Alessio Cargioli | Alessio Cargioli, Miguel Montesinos Ballester, Sonja Gantner, Emilio Gini, Mattias Beck, Jerome Faist | Quantum Walk Comb in a Dual Waveguide Quantum Cascade Laser | null | null | null | null | quant-ph physics.app-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ring quantum cascade lasers (QCLs) proved to be a versatile tool for generating tunable and stable frequency combs in the mid infrared range in the form of quantum walk combs. By homogeneously integrating a racetrack QCL with a passive waveguide, which lays on top of the active region plane and therefore can be designed to be fully independent from the laser geometry, we improve the light outcoupling from the ring by more than 2 orders of magnitude reaching a maximum output power of 120 mW. In addition, we show that it is possible to achieve quantum walk comb operation in the devices under analysis. Finally, we prove that we can change the light dispersion by tuning the parameters of the passive waveguide, with a direct impact on the behavior of the generated comb. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:28:29 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Cargioli",
"Alessio",
""
],
[
"Ballester",
"Miguel Montesinos",
""
],
[
"Gantner",
"Sonja",
""
],
[
"Gini",
"Emilio",
""
],
[
"Beck",
"Mattias",
""
],
[
"Faist",
"Jerome",
""
]
] |
2505.22248 | Armin Gie{\ss}ler | Armin Gie{\ss}ler, Felix Strehle, Jochen Illerhaus, and S\"oren Hohmann | Dynamic State-Feedback Control for LPV Systems: Ensuring Stability and LQR Performance | submitted to Transactions on Automatic Control | null | null | null | eess.SY cs.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we propose a novel dynamic state-feedback controller for polytopic linear parameter-varying (LPV) systems with constant input matrix. The controller employs a projected gradient flow method to continuously improve its control law and, under established conditions, converges to the optimal feedback gain of the corresponding linear quadratic regulator (LQR) problem associated with constant parameter trajectories. We derive conditions for quadratic stability, which can be verified via convex optimization, to ensure exponential stability of the LPV system even under arbitrarily fast parameter variations. Additionally, we provide sufficient conditions to guarantee the boundedness of the trajectories of the dynamic controller for any parameter trajectory and the convergence of its feedback gains to the optimal LQR gains for constant parameter trajectories. Furthermore, we show that the closed-loop system is asymptotically stable for constant parameter trajectories under these conditions. Simulation results demonstrate that the controller maintains stability and improves transient performance. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:30:09 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gießler",
"Armin",
""
],
[
"Strehle",
"Felix",
""
],
[
"Illerhaus",
"Jochen",
""
],
[
"Hohmann",
"Sören",
""
]
] |
2505.22249 | Yi Zhang | Cheng Hua, Arthur J. Swersey, Wenqian Xing, Yi Zhang | Optimizing Server Locations for Stochastic Emergency Service Systems | null | null | null | null | math.OC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a new model for solving the optimal server location problem in a stochastic system that accounts for unit availability, heterogeneity, and interdependencies. We show that this problem is NP-hard and derive both lower and upper bounds for the optimal solution by leveraging a special case of the classic $p$-Median problem. To overcome the computational challenges, we propose two Bayesian optimization approaches: (i) a parametric method that employs a sparse Bayesian linear model with a horseshoe prior (SparBL), and (ii) a non-parametric method based on a Gaussian process surrogate model with $p$-Median as mean prior (GP-$p$M). We prove that both algorithms achieve sublinear regret rates and converge to the optimal solution, with the parametric approach demonstrating particular effectiveness in high-dimensional settings. Numerical experiments and a case study using real-world data from St. Paul, Minnesota emergency response system show that our approaches consistently and efficiently identify optimal solutions, significantly outperforming the $p$-Median solution and other baselines. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:32:36 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hua",
"Cheng",
""
],
[
"Swersey",
"Arthur J.",
""
],
[
"Xing",
"Wenqian",
""
],
[
"Zhang",
"Yi",
""
]
] |
2505.22251 | Yuan Tseng | Yuan Tseng, Titouan Parcollet, Rogier van Dalen, Shucong Zhang, Sourav Bhattacharya | Evaluation of LLMs in Speech is Often Flawed: Test Set Contamination in Large Language Models for Speech Recognition | null | null | null | null | eess.AS cs.CL | http://creativecommons.org/licenses/by/4.0/ | Recent work suggests that large language models (LLMs) can improve performance of speech tasks compared to existing systems. To support their claims, results on LibriSpeech and Common Voice are often quoted. However, this work finds that a substantial amount of the LibriSpeech and Common Voice evaluation sets appear in public LLM pretraining corpora. This calls into question the reliability of findings drawn from these two datasets. To measure the impact of contamination, LLMs trained with or without contamination are compared, showing that a contaminated LLM is more likely to generate test sentences it has seen during training. Speech recognisers using contaminated LLMs shows only subtle differences in error rates, but assigns significantly higher probabilities to transcriptions seen during training. Results show that LLM outputs can be biased by tiny amounts of data contamination, highlighting the importance of evaluating LLM-based speech systems with held-out data. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:39:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Tseng",
"Yuan",
""
],
[
"Parcollet",
"Titouan",
""
],
[
"van Dalen",
"Rogier",
""
],
[
"Zhang",
"Shucong",
""
],
[
"Bhattacharya",
"Sourav",
""
]
] |
2505.22252 | Magdalena Proszewska | Magdalena Proszewska, Tomasz Danel, Dawid Rymarczyk | B-XAIC Dataset: Benchmarking Explainable AI for Graph Neural Networks Using Chemical Data | 26 pages, 16 figures, 5 tables | null | null | null | cs.LG cs.CE | http://creativecommons.org/licenses/by-sa/4.0/ | Understanding the reasoning behind deep learning model predictions is crucial in cheminformatics and drug discovery, where molecular design determines their properties. However, current evaluation frameworks for Explainable AI (XAI) in this domain often rely on artificial datasets or simplified tasks, employing data-derived metrics that fail to capture the complexity of real-world scenarios and lack a direct link to explanation faithfulness. To address this, we introduce B-XAIC, a novel benchmark constructed from real-world molecular data and diverse tasks with known ground-truth rationales for assigned labels. Through a comprehensive evaluation using B-XAIC, we reveal limitations of existing XAI methods for Graph Neural Networks (GNNs) in the molecular domain. This benchmark provides a valuable resource for gaining deeper insights into the faithfulness of XAI, facilitating the development of more reliable and interpretable models. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:40:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Proszewska",
"Magdalena",
""
],
[
"Danel",
"Tomasz",
""
],
[
"Rymarczyk",
"Dawid",
""
]
] |
2505.22253 | Jeffrey Galkowski | Yan-Long Fang and Jeffrey Galkowski | Surface plasmons in metamaterial cavities: Scattering by obstacles with negative wave speed | 59 pages and 4 figures | null | null | null | math.SP math-ph math.AP math.MP physics.optics | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study scattering by metamaterials with negative indices of refraction, which are known to support \emph{surface plasmons} -- long-lived states that are highly localized at the boundary of the cavity. This type of states has found uses in a variety of modern technologies. In this article, we study surface plasmons in the setting of non-trapping cavities; i.e. when all billiard trajectories outside the cavity escape to infinity. We characterize the indices of refraction which support surface plasmons, show that the corresponding resonances lie super-polynomially close to the real axis, describe the localization properties of the corresponding resonant states, and give an asymptotic formula for their number. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:41:07 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Fang",
"Yan-Long",
""
],
[
"Galkowski",
"Jeffrey",
""
]
] |
2505.22254 | Xiangxiang Dai | Xiangxiang Dai, Xiaowei Sun, Jinhang Zuo, Xutong Liu, John C.S. Lui | A Unified Online-Offline Framework for Co-Branding Campaign Recommendations | Accepted at the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2025 | null | 10.1145/3711896.3736824 | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Co-branding has become a vital strategy for businesses aiming to expand market reach within recommendation systems. However, identifying effective cross-industry partnerships remains challenging due to resource imbalances, uncertain brand willingness, and ever-changing market conditions. In this paper, we provide the first systematic study of this problem and propose a unified online-offline framework to enable co-branding recommendations. Our approach begins by constructing a bipartite graph linking ``initiating'' and ``target'' brands to quantify co-branding probabilities and assess market benefits. During the online learning phase, we dynamically update the graph in response to market feedback, while striking a balance between exploring new collaborations for long-term gains and exploiting established partnerships for immediate benefits. To address the high initial co-branding costs, our framework mitigates redundant exploration, thereby enhancing short-term performance while ensuring sustainable strategic growth. In the offline optimization phase, our framework consolidates the interests of multiple sub-brands under the same parent brand to maximize overall returns, avoid excessive investment in single sub-brands, and reduce unnecessary costs associated with over-prioritizing a single sub-brand. We present a theoretical analysis of our approach, establishing a highly nontrivial sublinear regret bound for online learning in the complex co-branding problem, and enhancing the approximation guarantee for the NP-hard offline budget allocation optimization. Experiments on both synthetic and real-world co-branding datasets demonstrate the practical effectiveness of our framework, with at least 12\% improvement. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:41:07 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Dai",
"Xiangxiang",
""
],
[
"Sun",
"Xiaowei",
""
],
[
"Zuo",
"Jinhang",
""
],
[
"Liu",
"Xutong",
""
],
[
"Lui",
"John C. S.",
""
]
] |
2505.22255 | Yaroslav Aksenov | Vadim Kurochkin, Yaroslav Aksenov, Daniil Laptev, Daniil Gavrilov and Nikita Balagansky | Train Sparse Autoencoders Efficiently by Utilizing Features Correlation | null | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | Sparse Autoencoders (SAEs) have demonstrated significant promise in interpreting the hidden states of language models by decomposing them into interpretable latent directions. However, training SAEs at scale remains challenging, especially when large dictionary sizes are used. While decoders can leverage sparse-aware kernels for efficiency, encoders still require computationally intensive linear operations with large output dimensions. To address this, we propose KronSAE, a novel architecture that factorizes the latent representation via Kronecker product decomposition, drastically reducing memory and computational overhead. Furthermore, we introduce mAND, a differentiable activation function approximating the binary AND operation, which improves interpretability and performance in our factorized framework. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:41:11 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Kurochkin",
"Vadim",
""
],
[
"Aksenov",
"Yaroslav",
""
],
[
"Laptev",
"Daniil",
""
],
[
"Gavrilov",
"Daniil",
""
],
[
"Balagansky",
"Nikita",
""
]
] |
2505.22256 | Yizhong Chen | Yizhong Chen and Zhibin Wang | Improvement of Solid-Fluid Interaction Scheme in Lattice Boltzmann Immiscible Pseudopotential Models | null | null | null | null | physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The pseudopotential model within the Lattice Boltzmann Method (LBM) framework has emerged as a prominent approach in computational fluid dynamics due to its dual strengths in physical intuitiveness and computational tractability. However, when modeling wettability phenomenon, existing solid-fluid interaction schemes exhibit persistent challenges in multi-component immiscible fluid systems, notably manifested through spurious velocity artifacts and unphysical mass-transfer boundary layers. This study presents an improved interaction scheme that preserves implementation simplicity while effectively mitigating these numerical artifacts. Furthermore, leveraging the enhanced isotropy characteristics of eighth-order discrete schemes, we develop a novel boundary treatment methodology addressing second-layer lattice data reconstruction at complex interfaces. To verify the universality of the proposed optimization scheme, four benchmark scenarios, including static contact angle measurement on cylindrical surface, droplet dynamics through confined geometries, immiscible displacement processes, and co-current flow in microchannels, are simulated to demonstrate the proposed scheme's capability. The results show that the improved scheme can well simulate various complex immiscible multiphase flows. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:42:05 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Chen",
"Yizhong",
""
],
[
"Wang",
"Zhibin",
""
]
] |
2505.22257 | Youssef Mroueh | Youssef Mroueh, Nicolas Dupuis, Brian Belgodere, Apoorva Nitsure, Mattia Rigotti, Kristjan Greenewald, Jiri Navratil, Jerret Ross, Jesus Rios | Revisiting Group Relative Policy Optimization: Insights into On-Policy and Off-Policy Training | null | null | null | null | cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | We revisit Group Relative Policy Optimization (GRPO) in both on-policy and off-policy optimization regimes. Our motivation comes from recent work on off-policy Proximal Policy Optimization (PPO), which improves training stability, sampling efficiency, and memory usage. In addition, a recent analysis of GRPO suggests that estimating the advantage function with off-policy samples could be beneficial. Building on these observations, we adapt GRPO to the off-policy setting. We show that both on-policy and off-policy GRPO objectives yield an improvement in the reward. This result motivates the use of clipped surrogate objectives in the off-policy version of GRPO. We then compare the empirical performance of reinforcement learning with verifiable rewards in post-training using both GRPO variants. Our results show that off-policy GRPO either significantly outperforms or performs on par with its on-policy counterpart. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:42:33 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Mroueh",
"Youssef",
""
],
[
"Dupuis",
"Nicolas",
""
],
[
"Belgodere",
"Brian",
""
],
[
"Nitsure",
"Apoorva",
""
],
[
"Rigotti",
"Mattia",
""
],
[
"Greenewald",
"Kristjan",
""
],
[
"Navratil",
"Jiri",
""
],
[
"Ross",
"Jerret",
""
],
[
"Rios",
"Jesus",
""
]
] |
2505.22258 | Benjamin Serfling | Benjamin Serfling, Hannes Reichert, Lorenzo Bayerlein, Konrad Doll and Kati Radkhah-Lens | LiDAR Based Semantic Perception for Forklifts in Outdoor Environments | null | null | null | null | cs.RO cs.CV cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this study, we present a novel LiDAR-based semantic segmentation framework tailored for autonomous forklifts operating in complex outdoor environments. Central to our approach is the integration of a dual LiDAR system, which combines forward-facing and downward-angled LiDAR sensors to enable comprehensive scene understanding, specifically tailored for industrial material handling tasks. The dual configuration improves the detection and segmentation of dynamic and static obstacles with high spatial precision. Using high-resolution 3D point clouds captured from two sensors, our method employs a lightweight yet robust approach that segments the point clouds into safety-critical instance classes such as pedestrians, vehicles, and forklifts, as well as environmental classes such as driveable ground, lanes, and buildings. Experimental validation demonstrates that our approach achieves high segmentation accuracy while satisfying strict runtime requirements, establishing its viability for safety-aware, fully autonomous forklift navigation in dynamic warehouse and yard environments. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:45:14 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Serfling",
"Benjamin",
""
],
[
"Reichert",
"Hannes",
""
],
[
"Bayerlein",
"Lorenzo",
""
],
[
"Doll",
"Konrad",
""
],
[
"Radkhah-Lens",
"Kati",
""
]
] |
2505.22259 | Kiyoon Jeong | Kiyoon Jeong, Jaehyuk Heo, Junyeong Son, Pilsung Kang | Domain Adaptation of Attention Heads for Zero-shot Anomaly Detection | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Zero-shot anomaly detection (ZSAD) in images is an approach that can detect anomalies without access to normal samples, which can be beneficial in various realistic scenarios where model training is not possible. However, existing ZSAD research has shown limitations by either not considering domain adaptation of general-purpose backbone models to anomaly detection domains or by implementing only partial adaptation to some model components. In this paper, we propose HeadCLIP to overcome these limitations by effectively adapting both text and image encoders to the domain. HeadCLIP generalizes the concepts of normality and abnormality through learnable prompts in the text encoder, and introduces learnable head weights to the image encoder to dynamically adjust the features held by each attention head according to domain characteristics. Additionally, we maximize the effect of domain adaptation by introducing a joint anomaly score that utilizes domain-adapted pixel-level information for image-level anomaly detection. Experimental results using multiple real datasets in both industrial and medical domains show that HeadCLIP outperforms existing ZSAD techniques at both pixel and image levels. In the industrial domain, improvements of up to 4.9%p in pixel-level mean anomaly detection score (mAD) and up to 3.0%p in image-level mAD were achieved, with similar improvements (3.2%p, 3.1%p) in the medical domain. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:45:51 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Jeong",
"Kiyoon",
""
],
[
"Heo",
"Jaehyuk",
""
],
[
"Son",
"Junyeong",
""
],
[
"Kang",
"Pilsung",
""
]
] |
2505.22260 | Philipp Maass | Alexander P. Antonov, Annika Vonhusen, Artem Ryabov, and Philipp Maass | Cluster sizes, particle displacements and currents in transport mediated by solitary cluster waves | 9 pages, 5 figures | null | null | null | cond-mat.stat-mech nlin.PS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In overdamped particle motion across periodic landscapes, solitary cluster waves can occur at high particle densities and lead to particle transport even in the absence of thermal noise. Here we show that for driven motion under a constant drag, the sum of all particle displacements per soliton equals one wavelength of the periodic potential. This unit displacement law is used to determine particle currents mediated by the solitons. We furthermore derive properties of clusters involved in the wave propagation as well as relations between cluster sizes and soliton numbers. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:46:43 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Antonov",
"Alexander P.",
""
],
[
"Vonhusen",
"Annika",
""
],
[
"Ryabov",
"Artem",
""
],
[
"Maass",
"Philipp",
""
]
] |
2505.22261 | Z. A. Jia | Qiang Jia, Zhian Jia | Subsystem Symmetry-Protected Topological Phases from Subsystem SymTFT of 2-Foliated Exotic Tensor Gauge Theory | v1: 59 pages | null | null | null | cond-mat.str-el hep-th math-ph math.MP quant-ph | http://creativecommons.org/licenses/by/4.0/ | Symmetry topological field theory (SymTFT), or topological holography, posits a correspondence between symmetries in a $d$-dimensional theory and topological order in a $(d+1)$-dimensional theory. In this work, we extend this framework to subsystem symmetries and develop subsystem SymTFT as a systematic tool to characterize and classify subsystem symmetry-protected topological (SSPT) phases. For $(2+1)$D gapped phases, we introduce a 2-foliated $(3+1)$D exotic tensor gauge theory (which is equivalent to 2-foliated $(3+1)$D BF theory via exotic duality) as the subsystem SymTFT and systematically analyze its topological boundary conditions and linearly rigid subsystem symmetries. Taking subsystem symmetry groups $G = \mathbb{Z}_N$ and $G=\mathbb{Z}_N \times \mathbb{Z}_M$ as examples, we demonstrate how to recover the classification scheme $\mathcal{C}[G] = H^{2}(G^{\times 2}, U(1)) / \left( H^2(G, U(1)) \right)^3$, which was previously derived by examining topological invariant under linear subsystem-symmetric local unitary transformations in the lattice Hamiltonian formalism. To illustrate the correspondence between field-theoretic and lattice descriptions, we further analyze $\mathbb{Z}_2 \times \mathbb{Z}_2$ and $\mathbb{Z}_N \times \mathbb{Z}_M$ cluster state models as concrete examples. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:47:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Jia",
"Qiang",
""
],
[
"Jia",
"Zhian",
""
]
] |
2505.22262 | Md Aquib Molla | Md Aquib Molla, Sanchari Goswami and Parongama Sen | Forager with Inertia : Better for Survival? | 10 pages, 10 figures | null | null | null | cond-mat.stat-mech physics.bio-ph physics.soc-ph | http://creativecommons.org/licenses/by/4.0/ | We study the fate of a forager who searches for food performing simple random walk on lattices. The forager consumes the available food on the site it visits and leaves it depleted but can survive up to $S$ steps without food. We introduce the concept of inertia in the dynamics which allows the forager to rest with probability $p$ upon consumption of food. The parameter $p$ significantly affects the lifetime of the forager, showing that the inertia can be beneficial for the forager for chosen parameter values. The study of various other quantities reveals interesting scaling behavior with $p$ and also departure from usual diffusive behavior for $0.5 < p < 1$. In addition to numerical approach, the problem has also been studied with analytical approach in one dimension and the results agree reasonably well. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:50:01 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Molla",
"Md Aquib",
""
],
[
"Goswami",
"Sanchari",
""
],
[
"Sen",
"Parongama",
""
]
] |
2505.22263 | Ansar Safin | David A. Gabrielyan, Dmitry A. Volkov, Tatyana V. Bogdanova, Kristina D. Samoylenko, Anton V. Matasov, Ansar R. Safin, Dmitry V. Kalyabin, Alexey A. Klimov, Leonid M. Krutyansky, Vladimir L. Preobrazhensky, Sergey A. Nikitov | Ultrasonic spin pumping in the antiferromagnetic acoustic resonator $\alpha-\text{Fe}_2\text{O}_3$ | 19 pages, 6 figures, 1 table | null | null | null | cond-mat.other physics.app-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advances in magnon spintronics have ignited interest in the interactions between the spin and elastic subsystems of magnetic materials. These interactions suggest a dynamic connection between collective excitations of spins, quantized as magnons, and elastic waves generated by perturbations in the crystal lattice, quantized as phonons. Both magnons and their associated magnon-phonon excitations can act as sources of spin pumping from magnetic materials into non-magnetic metals. Although a considerable body of research has focused on spin pumping via elastic waves in ferromagnets, similar investigations involving antiferromagnets have yet to be undertaken. In this work, we experimentally demonstrate for the first time the feasibility of generating spin currents at ultrasonic frequencies of acoustic resonance in antiferromagnetic crystal hematite $\alpha-\text{Fe}_2\text{O}_3$ at room temperature. We provide both theoretical and experimental evidence that, due to strong magnetoelastic coupling, acoustic vibrations in hematite induce significant variable deviations in magnetization, resulting in spin accumulation at the antiferromagnet-normal metal interface, which in turn leads to the generation of spin and charge currents in the metal. Charge currents arising from the inverse spin Hall effect can be measured using the same methodology employed under high-frequency spin pumping conditions at the resonances of the magnetic subsystem itself. Moreover, the acoustic resonance in hematite is significantly more pronounced (by hundreds or even thousands of times) than in other quasiferromagnetic or antiferromagnetic systems, enabling the attainment of extremely large amplitudes of magnetic oscillations for spin pumping. This research highlights the new approach of utilizing acoustic spin pumping to manipulate spin currents in magnetic materials, particularly antiferromagnets. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:50:18 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Gabrielyan",
"David A.",
""
],
[
"Volkov",
"Dmitry A.",
""
],
[
"Bogdanova",
"Tatyana V.",
""
],
[
"Samoylenko",
"Kristina D.",
""
],
[
"Matasov",
"Anton V.",
""
],
[
"Safin",
"Ansar R.",
""
],
[
"Kalyabin",
"Dmitry V.",
""
],
[
"Klimov",
"Alexey A.",
""
],
[
"Krutyansky",
"Leonid M.",
""
],
[
"Preobrazhensky",
"Vladimir L.",
""
],
[
"Nikitov",
"Sergey A.",
""
]
] |
2505.22264 | Maximiliano Hormazabal | Maximiliano Hormaz\'abal Lagos,\'Alvaro Bueno Saez, H\'ector Cerezo-Costas, Pedro Alonso Doval, Jorge Alcalde Vesteiro | MRT at SemEval-2025 Task 8: Maximizing Recovery from Tables with Multiple Steps | 7 pages, 6 tables | null | null | null | cs.CL cs.AI cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper we expose our approach to solve the \textit{SemEval 2025 Task 8: Question-Answering over Tabular Data} challenge. Our strategy leverages Python code generation with LLMs to interact with the table and get the answer to the questions. The process is composed of multiple steps: understanding the content of the table, generating natural language instructions in the form of steps to follow in order to get the answer, translating these instructions to code, running it and handling potential errors or exceptions. These steps use open source LLMs and fine grained optimized prompts for each task (step). With this approach, we achieved a score of $70.50\%$ for subtask 1. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:50:22 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lagos",
"Maximiliano Hormazábal",
""
],
[
"Saez",
"Álvaro Bueno",
""
],
[
"Cerezo-Costas",
"Héctor",
""
],
[
"Doval",
"Pedro Alonso",
""
],
[
"Vesteiro",
"Jorge Alcalde",
""
]
] |
2505.22265 | Satyajeet Patil | Satyajeet Patil, Sebastian T\"opfer, Rajshree Swarnkar, Sergio Tovar-Perez, Jonas Moos, Jorge Fuenzalida, and Markus Gr\"afe | Advances in Position-Momentum Entanglement: A Versatile Tool for Quantum Technologies | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Position-momentum entanglement is a versatile high-dimensional resource in quantum optics. From fundamental tests of reality, to application in quantum technologies, spatial entanglement has had an increasing growth in recent years. In this review, we explore these advances, starting from the generation of spatial entanglement, followed by the different types of measurements for quantifying the entanglement, and finishing with different quantum-based applications. We conclude the review with a discussion and future perspectives on the field. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:51:21 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Patil",
"Satyajeet",
""
],
[
"Töpfer",
"Sebastian",
""
],
[
"Swarnkar",
"Rajshree",
""
],
[
"Tovar-Perez",
"Sergio",
""
],
[
"Moos",
"Jonas",
""
],
[
"Fuenzalida",
"Jorge",
""
],
[
"Gräfe",
"Markus",
""
]
] |
2505.22266 | Jialin Yan | Jialin Yan, Yu Cheng, Zhaoxia Yin, Xinpeng Zhang, Shilin Wang, Tanfeng Sun, Xinghao Jiang | FGS-Audio: Fixed-Decoder Framework for Audio Steganography with Adversarial Perturbation Generation | null | null | null | null | cs.SD cs.MM eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rapid development of Artificial Intelligence Generated Content (AIGC) has made high-fidelity generated audio widely available across the Internet, offering an abundant and versatile source of cover signals for covert communication. Driven by advances in deep learning, current audio steganography frameworks are mainly based on encoding-decoding network architectures. While these methods greatly improve the security of audio steganography, they typically employ elaborate training workflows and rely on extensive pre-trained models. To address the aforementioned issues, this paper pioneers a Fixed-Decoder Framework for Audio Steganography with Adversarial Perturbation Generation (FGS-Audio). The adversarial perturbations that carry secret information are embedded into cover audio to generate stego audio. The receiver only needs to share the structure and weights of the fixed decoding network to accurately extract the secret information from the stego audio, thus eliminating the reliance on large pre-trained models. In FGS-Audio, we propose an audio Adversarial Perturbation Generation (APG) strategy and design a lightweight fixed decoder. The fixed decoder guarantees reliable extraction of the hidden message, while the adversarial perturbations are optimized to keep the stego audio perceptually and statistically close to the cover audio, thereby improving resistance to steganalysis. The experimental results show that the method exhibits excellent anti-steganalysis performance under different relative payloads, outperforming existing SOTA approaches. In terms of stego audio quality, FGS-Audio achieves an average PSNR improvement of over 10 dB compared to SOTA method. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:54:09 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Yan",
"Jialin",
""
],
[
"Cheng",
"Yu",
""
],
[
"Yin",
"Zhaoxia",
""
],
[
"Zhang",
"Xinpeng",
""
],
[
"Wang",
"Shilin",
""
],
[
"Sun",
"Tanfeng",
""
],
[
"Jiang",
"Xinghao",
""
]
] |
2505.22267 | Ilan Bouquet | Ilan Bouquet, Jiang Cao, and Mathieu Luisier | Simulation of single hole spin qubit in strained triangular FinFET quantum devices | null | null | null | null | quant-ph cond-mat.mes-hall | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Using an in-house Schroedinger-Poisson (SP) solver, we investigate the creation of a single hole spin qubit inside a triple-gate triangular silicon fin field effect transistor (Si FinFET) quantum device similar to experimental structures. The gate induced formation of the required quantum dot (QD) is monitored based on the Luttinger-Kohn 6x6 kp method accounting for magnetic fields and strain to determine the qubit ground state. Strain arises from the inhomogeneous contraction of the different FinFET components when they are cooled down to cryogenic temperatures. It leads to a renormalization of the qubit energy levels, thus impacting both the heavy-hole (HH) and light-hole (LH) populations as well as their mixing. The dot length, band mixing, g-factor, and Larmor/Rabi frequencies of the considered device are extracted. In particular, we show that these metrics exhibit strong strain-dependent variations of their magnitude, thus underlying the importance of including realistic thermal contraction scenarios when modeling hole spin qubits. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:54:20 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bouquet",
"Ilan",
""
],
[
"Cao",
"Jiang",
""
],
[
"Luisier",
"Mathieu",
""
]
] |
2505.22268 | Darshan Raju | Darshan Raju, Mahinder Ramdin, and Thijs J.H. Vlugt | Thermophysical Properties and Phase Behavior of CO2 with Impurities: Insight from Molecular Simulations | null | null | null | null | physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | Experimentally determining thermophysical properties for various compositions commonly found in CO2 transportation systems is extremely challenging. To overcome this challenge, we performed Monte Carlo (MC) and molecular dynamics (MD) simulations of CO2 rich mixtures to compute thermophysical properties such as densities, thermal expansion coefficients, isothermal compressibilities, heat capacities, Joule-Thomson coefficients, speed of sound, and viscosities at temperatures (235-313) K and pressures (20-200) bar. We computed thermophysical properties of pure CO2 and CO2 rich mixtures with N2, Ar, H2, and CH4 as impurities (1-10) mol% and showed good agreement with available equations of state (EoS). We showed that impurities decrease the values of thermal expansion coefficients, isothermal compressibilities, heat capacities, and Joule-Thomson coefficients in the gas phase, while these values increase in the liquid and supercritical phases. In contrast, impurities increase the value of speed of sound in the gas phase and decrease it in the liquid and supercritical phases. We present an extensive data set of thermophysical properties for CO2 rich mixtures with various impurities, which will help to design the safe and efficient operation of CO2 transportation systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:54:55 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Raju",
"Darshan",
""
],
[
"Ramdin",
"Mahinder",
""
],
[
"Vlugt",
"Thijs J. H.",
""
]
] |
2505.22269 | Thomas Burger | Thomas SJ Burger, Amir Shahhosseini, Rodolphe Sepulchre | A memristive model of spatio-temporal excitability | 7 pages, 9 figures, submitted for CDC 2025 | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | This paper introduces a model of excitability that unifies the mechanism of an important neuronal property both in time and in space. As a starting point, we revisit both a key model of temporal excitability, proposed by Hodgkin and Huxley, and a key model of spatial excitability, proposed by Amari. We then propose a novel model that captures the temporal and spatial properties of both models. Our aim is to regard neuronal excitability as a property across scales, and to explore the benefits of modeling excitability with one and the same mechanism, whether at the cellular or the population level. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:55:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Burger",
"Thomas SJ",
""
],
[
"Shahhosseini",
"Amir",
""
],
[
"Sepulchre",
"Rodolphe",
""
]
] |
2505.22270 | Adam B. Cahaya | Adam B. Cahaya | Momentum and spin dynamics of photon in linear dielectric medium | null | null | null | null | physics.optics quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The momentum of light in dielectric media has been a century-long controversy that continues to attract significant interest. In a linear dielectric medium with refractive index $n$, the momentum is predicted to be smaller by a factor of $n$ according to Max Abraham, and larger by the same factor according to Hermann Minkowski. By studying the coupled dynamics of photons and dipoles in a dielectric medium, we show that the change in momentum of the dipole corresponds to the Abraham momentum. On the other hand, the Minkowski momentum determines the direction of propagation and the precession of the photon spin, which is described by an SO(3) rotation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:57:10 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Cahaya",
"Adam B.",
""
]
] |
2505.22271 | Yongcan Yu | Yongcan Yu, Yanbo Wang, Ran He, Jian Liang | Test-Time Immunization: A Universal Defense Framework Against Jailbreaks for (Multimodal) Large Language Models | Under Review | null | null | null | cs.CR cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While (multimodal) large language models (LLMs) have attracted widespread attention due to their exceptional capabilities, they remain vulnerable to jailbreak attacks. Various defense methods are proposed to defend against jailbreak attacks, however, they are often tailored to specific types of jailbreak attacks, limiting their effectiveness against diverse adversarial strategies. For instance, rephrasing-based defenses are effective against text adversarial jailbreaks but fail to counteract image-based attacks. To overcome these limitations, we propose a universal defense framework, termed Test-time IMmunization (TIM), which can adaptively defend against various jailbreak attacks in a self-evolving way. Specifically, TIM initially trains a gist token for efficient detection, which it subsequently applies to detect jailbreak activities during inference. When jailbreak attempts are identified, TIM implements safety fine-tuning using the detected jailbreak instructions paired with refusal answers. Furthermore, to mitigate potential performance degradation in the detector caused by parameter updates during safety fine-tuning, we decouple the fine-tuning process from the detection module. Extensive experiments on both LLMs and multimodal LLMs demonstrate the efficacy of TIM. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:57:46 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Yu",
"Yongcan",
""
],
[
"Wang",
"Yanbo",
""
],
[
"He",
"Ran",
""
],
[
"Liang",
"Jian",
""
]
] |
2505.22272 | Igor V. Nikolaev | Igor V. Nikolaev | Addendum to "Measured foliations and Hilbert 12th problem" | 4 pages, 1 table | null | null | null | math.NT math.OA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study numerical examples of the abelian extensions of the real quadratic number fields based on the results in Acta Mathematica Vietnamica 48 (2023), 271-281 (arXiv:0804.0057) | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:58:32 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Nikolaev",
"Igor V.",
""
]
] |
2505.22273 | Shohei Higashiyama | Shohei Higashiyama and Masao Utiyama | Comprehensive Evaluation on Lexical Normalization: Boundary-Aware Approaches for Unsegmented Languages | 23 pages | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Lexical normalization research has sought to tackle the challenge of processing informal expressions in user-generated text, yet the absence of comprehensive evaluations leaves it unclear which methods excel across multiple perspectives. Focusing on unsegmented languages, we make three key contributions: (1) creating a large-scale, multi-domain Japanese normalization dataset, (2) developing normalization methods based on state-of-the-art pretrained models, and (3) conducting experiments across multiple evaluation perspectives. Our experiments show that both encoder-only and decoder-only approaches achieve promising results in both accuracy and efficiency. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:02:45 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Higashiyama",
"Shohei",
""
],
[
"Utiyama",
"Masao",
""
]
] |
2505.22274 | Bin Cheng | Yanwei Cui, Zenglin Liu, Qin Liu, Junlin Xiong, Yongqin Xie, Yudi Dai, Ji Zhou, Lizheng Wang, Hanyan Fang, Haiwen Liu, Shi-Jun Liang, Bin Cheng, and Feng Miao | Dimensionality-Driven Anomalous Metallic State with Zero-field Nonreciprocal Transport in Layered Ising Superconductors | null | null | null | null | cond-mat.supr-con cond-mat.mes-hall | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The anomalous metal state (AMS), observed in failed superconductors, provides insights into superconductivity and quantum criticality, with studies revealing unconventional quantum phases like the Bose metal. Recently, layered transition metal dichalcogenide (TMD) superconductors approaching the two-dimensional limit have garnered significant attention for the enhanced phase fluctuations and electronic correlations. Investigating AMS in these systems, particularly in the absence of an external magnetic field, could offer valuable insights into the dimensionality-driven emergence of exotic quantum phenomena, including triplet Cooper pairing, phase fluctuation dynamics, and especially the recently discovered field-free superconducting diode effects. However, the field-free AMS has yet to be observed in TMD superconductors. Here, we report the dimensionality-tunable AMS near the superconducting quantum phase transitions in a layered TMD superconductor 2H-Ta2S3Se. In samples with thicknesses below 10 nm, we demonstrate magnetic field-driven AMS under external magnetic field, characterized by the vanishing of the Hall resistance and the presence of finite longitudinal resistance. Remarkably, an unexpected zero-field AMS emerges as the sample thickness is reduced to 3 nm. This AMS aligns well with the quantum vortex creep model and exhibits non-reciprocal transport behaviors, suggesting the onset of spontaneous time-reversal symmetry breaking accompanied by vortex motion as the system approaches the two-dimensional limit. Our findings open new avenues for exploring dimensionality-driven exotic superconducting quantum critical phases, and pave the way for a deeper understanding of zero-field superconducting diode effects. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:04:59 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Cui",
"Yanwei",
""
],
[
"Liu",
"Zenglin",
""
],
[
"Liu",
"Qin",
""
],
[
"Xiong",
"Junlin",
""
],
[
"Xie",
"Yongqin",
""
],
[
"Dai",
"Yudi",
""
],
[
"Zhou",
"Ji",
""
],
[
"Wang",
"Lizheng",
""
],
[
"Fang",
"Hanyan",
""
],
[
"Liu",
"Haiwen",
""
],
[
"Liang",
"Shi-Jun",
""
],
[
"Cheng",
"Bin",
""
],
[
"Miao",
"Feng",
""
]
] |
2505.22275 | Alexander Hagg | Alexander Hagg, Adam Gaier, Dominik Wilde, Alexander Asteroth, Holger Foysi and Dirk Reith | Full Domain Analysis in Fluid Dynamics | null | null | null | null | cs.LG cs.NE | http://creativecommons.org/licenses/by/4.0/ | Novel techniques in evolutionary optimization, simulation and machine learning allow for a broad analysis of domains like fluid dynamics, in which computation is expensive and flow behavior is complex. Under the term of full domain analysis we understand the ability to efficiently determine the full space of solutions in a problem domain, and analyze the behavior of those solutions in an accessible and interactive manner. The goal of full domain analysis is to deepen our understanding of domains by generating many examples of flow, their diversification, optimization and analysis. We define a formal model for full domain analysis, its current state of the art, and requirements of subcomponents. Finally, an example is given to show what we can learn by using full domain analysis. Full domain analysis, rooted in optimization and machine learning, can be a helpful tool in understanding complex systems in computational physics and beyond. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:06:48 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Hagg",
"Alexander",
""
],
[
"Gaier",
"Adam",
""
],
[
"Wilde",
"Dominik",
""
],
[
"Asteroth",
"Alexander",
""
],
[
"Foysi",
"Holger",
""
],
[
"Reith",
"Dirk",
""
]
] |
2505.22276 | Mohammed Alghadeer | Mohammed Alghadeer, Shuxiang Cao, Simone D Fasciati, Michele Piscitelli, Paul C. Gow, James C. Gates, Mustafa Bakr, and Peter J. Leek | Low Crosstalk in a Scalable Superconducting Quantum Lattice | arXiv admin note: text overlap with arXiv:2501.15059 | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Superconducting quantum circuits are a key platform for advancing quantum information processing and simulation. Scaling efforts currently encounter challenges such as Josephson-junction fabrication yield, design frequency targeting, and crosstalk arising both from spurious microwave modes and intrinsic interactions between qubits. We demonstrate a scalable 4x4 square lattice with low crosstalk, comprising 16 fixed-frequency transmon qubits with nearest-neighbor capacitive coupling that is implemented in a tileable, 3D-integrated circuit architecture with off-chip inductive shunting to mitigate spurious enclosure modes. We report on the design and comprehensive characterization, and show that our implementation achieves targeted device parameters with very low frequency spreads and simultaneous single-qubit gate errors across the device. Our results provide a promising pathway toward a scalable, low-crosstalk superconducting lattice topology with high qubit connectivity for quantum error correction and simulation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:07:51 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Alghadeer",
"Mohammed",
""
],
[
"Cao",
"Shuxiang",
""
],
[
"Fasciati",
"Simone D",
""
],
[
"Piscitelli",
"Michele",
""
],
[
"Gow",
"Paul C.",
""
],
[
"Gates",
"James C.",
""
],
[
"Bakr",
"Mustafa",
""
],
[
"Leek",
"Peter J.",
""
]
] |
2505.22277 | Aggeliki Chalki | Luca Aceto, Antonis Achilleos, Aggeliki Chalki, Anna Ingolfsdottir | The complexity of deciding characteristic formulae modulo nested simulation | null | null | null | null | cs.LO | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper studies the complexity of determining whether a formula in the modal logics characterizing the nested-simulation semantics is characteristic for some process, which is equivalent to determining whether the formula is satisfiable and prime. The main results are that the problem of determining whether a formula is prime in the modal logic characterizing the 2-nested-simulation preorder is CoNP-complete and is PSPACE-complete in the case of the n-nested-simulation preorder, when n>= 3. This establishes that deciding characteristic formulae for the n-nested simulation semantics, n>= 3, is PSPACE-complete. In the case of the 2-nested simulation semantics, that problem lies in the complexity class DP, which consists of languages that can be expressed as the intersection of one language in NP and of one in CoNP. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:11:33 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Aceto",
"Luca",
""
],
[
"Achilleos",
"Antonis",
""
],
[
"Chalki",
"Aggeliki",
""
],
[
"Ingolfsdottir",
"Anna",
""
]
] |
2505.22278 | Afroja Parvin | Afroja Parvin, Giovanni Samaey, Julian Koellermeier | A Coupled Hydro-Morphodynamic Model for Sediment Transport using the Moment Approach | 42 pages, 11 figures | null | null | null | math.NA cs.NA physics.geo-ph | http://creativecommons.org/licenses/by/4.0/ | Sediment transport is crucial in the hydro-morphodynamic evolution of free surface flows in shallow water environments, which is typically modeled under the shallow water assumption. In classical shallow water modeling for sediment transport, the vertical structure of the flow is collapsed into a depth-averaged and near-bed velocity, usually reconstructed empirically, e.g., using a parameterized logarithmic profile. In practice, large variations from such empirical profiles can occur. It is therefore essential to resolve the vertical structure of the velocity profile within the shallow water framework to better approximate near-bed velocity. This study introduces a model and simulations that incorporate vertical velocity variations and bottom erosion-deposition effects in sediment transport, providing a computationally efficient framework for predicting sediment dynamics in shallow water environments. We employ the so-called moment model approach for the velocity variation, which considers a polynomial expansion of the horizontal velocity in the scaled vertical direction. This allows the use of a complex velocity profile with an extended set of variables determined by the polynomial basis coefficients, resolving the vertical structure as part of the solution. The extended model comprises four components: (1) the standard shallow water equations; (2) moment equations governing evolution of the basis coefficients; (3) an evolution equation for sediment concentration; and (4) a transport equation for the bed. This enables a coupled model for bedload and suspended load transport. We use a hyperbolic regularization technique to ensure model stability and realistic eigenvalues. Several numerical tests, including dam-break cases with and without wet/dry fronts, validate our results against laboratory data. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:13:43 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Parvin",
"Afroja",
""
],
[
"Samaey",
"Giovanni",
""
],
[
"Koellermeier",
"Julian",
""
]
] |
2505.22279 | Wenjun Lu Mr. | Wenjun Lu, Haodong Chen, Anqi Yi, Yuk Ying Chung, Zhiyong Wang, Kun Hu | Learning Fine-Grained Geometry for Sparse-View Splatting via Cascade Depth Loss | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Novel view synthesis is a fundamental task in 3D computer vision that aims to reconstruct realistic images from a set of posed input views. However, reconstruction quality degrades significantly under sparse-view conditions due to limited geometric cues. Existing methods, such as Neural Radiance Fields (NeRF) and the more recent 3D Gaussian Splatting (3DGS), often suffer from blurred details and structural artifacts when trained with insufficient views. Recent works have identified the quality of rendered depth as a key factor in mitigating these artifacts, as it directly affects geometric accuracy and view consistency. In this paper, we address these challenges by introducing Hierarchical Depth-Guided Splatting (HDGS), a depth supervision framework that progressively refines geometry from coarse to fine levels. Central to HDGS is a novel Cascade Pearson Correlation Loss (CPCL), which aligns rendered and estimated monocular depths across multiple spatial scales. By enforcing multi-scale depth consistency, our method substantially improves structural fidelity in sparse-view scenarios. Extensive experiments on the LLFF and DTU benchmarks demonstrate that HDGS achieves state-of-the-art performance under sparse-view settings while maintaining efficient and high-quality rendering | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:16:42 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lu",
"Wenjun",
""
],
[
"Chen",
"Haodong",
""
],
[
"Yi",
"Anqi",
""
],
[
"Chung",
"Yuk Ying",
""
],
[
"Wang",
"Zhiyong",
""
],
[
"Hu",
"Kun",
""
]
] |
2505.22280 | Zihan Xu | Zihan Xu, Haotian Ma, Gongbo Zhang, Yihao Ding, Chunhua Weng, Yifan Peng | Natural Language Processing in Support of Evidence-based Medicine: A Scoping Review | Accepted by ACL 2025 Findings | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Evidence-based medicine (EBM) is at the forefront of modern healthcare, emphasizing the use of the best available scientific evidence to guide clinical decisions. Due to the sheer volume and rapid growth of medical literature and the high cost of curation, there is a critical need to investigate Natural Language Processing (NLP) methods to identify, appraise, synthesize, summarize, and disseminate evidence in EBM. This survey presents an in-depth review of 129 research studies on leveraging NLP for EBM, illustrating its pivotal role in enhancing clinical decision-making processes. The paper systematically explores how NLP supports the five fundamental steps of EBM -- Ask, Acquire, Appraise, Apply, and Assess. The review not only identifies current limitations within the field but also proposes directions for future research, emphasizing the potential for NLP to revolutionize EBM by refining evidence extraction, evidence synthesis, appraisal, summarization, enhancing data comprehensibility, and facilitating a more efficient clinical workflow. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:17:01 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Xu",
"Zihan",
""
],
[
"Ma",
"Haotian",
""
],
[
"Zhang",
"Gongbo",
""
],
[
"Ding",
"Yihao",
""
],
[
"Weng",
"Chunhua",
""
],
[
"Peng",
"Yifan",
""
]
] |
2505.22281 | Philipp Schmidt-Wellenburg | M. Athanasakis-Kaklamanakis, M. Au, R. Berger, S. Degenkolb, J. De Vries, S. Hoekstra, A. Keshavarzi, N. Neri, D. Ries, P. Schmidt-Wellenburg, and M. Tarbutt (Imperial College London, United Kingdom) | Community input to the European Strategy on particle physics: Searches for Permanent Electric Dipole Moments | 14 pages, 3 figures, 2 tables, Input to the EPPSU 2025 | null | null | null | hep-ex nucl-ex | http://creativecommons.org/licenses/by/4.0/ | Searches for electric dipole moments (EDMs) in fundamental particles and quantum systems with spin are pivotal experiments at the intersection of low-energy and high-precision particle physics. These investigations offer a complementary pathway to uncovering new physics beyond the Standard Model, parallel to high-energy collider searches. EDM experiments are among the most sensitive probes for detecting non-standard time-reversal (T) symmetry violations and, via the CPT theorem, CP-violation (CPV). Current EDM measurements test new physics at mass scales in or above the 10TeV to 100TeV range. This community input to the European Particle Physics Strategy Update highlights the status of the field, and describes challenges and opportunities in Europe. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:17:15 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Athanasakis-Kaklamanakis",
"M.",
"",
"Imperial College London, United Kingdom"
],
[
"Au",
"M.",
"",
"Imperial College London, United Kingdom"
],
[
"Berger",
"R.",
"",
"Imperial College London, United Kingdom"
],
[
"Degenkolb",
"S.",
"",
"Imperial College London, United Kingdom"
],
[
"De Vries",
"J.",
"",
"Imperial College London, United Kingdom"
],
[
"Hoekstra",
"S.",
"",
"Imperial College London, United Kingdom"
],
[
"Keshavarzi",
"A.",
"",
"Imperial College London, United Kingdom"
],
[
"Neri",
"N.",
"",
"Imperial College London, United Kingdom"
],
[
"Ries",
"D.",
"",
"Imperial College London, United Kingdom"
],
[
"Schmidt-Wellenburg",
"P.",
"",
"Imperial College London, United Kingdom"
],
[
"Tarbutt",
"M.",
"",
"Imperial College London, United Kingdom"
]
] |
2505.22282 | Ken'ichi Yoshida | Ken'ichi Yoshida | Uniqueness of free 2-periodicities of links | 10 pages | null | null | null | math.GT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that if two links in the real projective 3-space $\mathbb{RP}^{3}$ have isotopic preimages in the 3-sphere $S^{3}$ by the double covering map, then they are themselves isotopic in $\mathbb{RP}^{3}$. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:17:32 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Yoshida",
"Ken'ichi",
""
]
] |
2505.22283 | Julia Ruohotie | Julia Ruohotie, Simon Good, Christian M\"ostl, Emilia Kilpua | Intermittency in Interplanetary Coronal Mass Ejections Observed by Parker Solar Probe and Solar Orbiter | 12 pages, 5 figures; accepted for publication in the Astrophysical Journal Letters 2025 May 27 | null | null | null | astro-ph.SR physics.plasm-ph physics.space-ph | http://creativecommons.org/licenses/by/4.0/ | Intermittency has been studied extensively in the fast and slow solar winds but to a far lesser extent in interplanetary coronal mass ejections (ICMEs). While ICMEs are often characterized by their relatively smooth, large-scale magnetic flux rope structures, a spectrum of fluctuations is nonetheless present at smaller scales. We have examined kurtosis and its scaling exponents at magnetohydrodynamic inertial scales in 49 ICMEs observed between 0.25 and 1 au by Parker Solar Probe and Solar Orbiter, and compared the results to those obtained for the ICME sheath regions and ambient solar wind intervals. Kurtosis behaves similarly in all intervals studied and presents a universal behavior typical of intermittent time series. The ICMEs displayed a radially invariant level of intermittency, suggesting that they are relatively static, well-developed turbulent environments. In the sheath regions, the level of intermittency increased with distance, indicating that the turbulence is not yet fully developed at small heliocentric distances. In addition to intermittent fluctuations related to turbulence, the sheath regions may possess a population of non-turbulent structures that increase the absolute value of kurtosis. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:19:50 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Ruohotie",
"Julia",
""
],
[
"Good",
"Simon",
""
],
[
"Möstl",
"Christian",
""
],
[
"Kilpua",
"Emilia",
""
]
] |
2505.22284 | Junyu Fan | Junyu Fan, Chuanlin Liao, Yi Lin | From Controlled Scenarios to Real-World: Cross-Domain Degradation Pattern Matching for All-in-One Image Restoration | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As a fundamental imaging task, All-in-One Image Restoration (AiOIR) aims to achieve image restoration caused by multiple degradation patterns via a single model with unified parameters. Although existing AiOIR approaches obtain promising performance in closed and controlled scenarios, they still suffered from considerable performance reduction in real-world scenarios since the gap of data distributions between the training samples (source domain) and real-world test samples (target domain) can lead inferior degradation awareness ability. To address this issue, a Unified Domain-Adaptive Image Restoration (UDAIR) framework is proposed to effectively achieve AiOIR by leveraging the learned knowledge from source domain to target domain. To improve the degradation identification, a codebook is designed to learn a group of discrete embeddings to denote the degradation patterns, and the cross-sample contrastive learning mechanism is further proposed to capture shared features from different samples of certain degradation. To bridge the data gap, a domain adaptation strategy is proposed to build the feature projection between the source and target domains by dynamically aligning their codebook embeddings, and a correlation alignment-based test-time adaptation mechanism is designed to fine-tune the alignment discrepancies by tightening the degradation embeddings to the corresponding cluster center in the source domain. Experimental results on 10 open-source datasets demonstrate that UDAIR achieves new state-of-the-art performance for the AiOIR task. Most importantly, the feature cluster validate the degradation identification under unknown conditions, and qualitative comparisons showcase robust generalization to real-world scenarios. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:22:00 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Fan",
"Junyu",
""
],
[
"Liao",
"Chuanlin",
""
],
[
"Lin",
"Yi",
""
]
] |
2505.22285 | Marco Bondi | M. Bondi, I. Prandoni, M. Magliocchetti, L. Bisigello, M. Bonato, M. Giulietti, R. Scaramella, G. Brunetti, F. Vitello | Sub-arcsecond-resolution LOFAR observations of bright sub-millimetre galaxies in the North Ecliptic Pole field | Accepted for publication in Astronomy & Astrophysics. 11 pages, 6 figures | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Bright SMGs contribute significantly to the star formation rate (SFR) density (20-50\%) and stellar mass density ($\sim$ 30-50\%) at $z=$ 2-4 with SFRs$\ge 1000$ M$_\odot$,yr$^{-1}$ and stellar masses of $\sim 10^{11}$-$10^{12}$ M$_\odot$. The number of bright SMGs with such high SFRs is hard to reconcile with the standard models of galaxy formation and evolution. In this paper we provide evidence that, in a small sample of 12 bright SMGs, the SFRs derived from spectral energy distribution (SED) fitting are significantly higher than those obtained using low-frequency radio emission as a proxy for star formation. Using the International LOFAR Telescope (ILT), which allows imaging at 144 MHz with sub-arcsecond angular resolution, we have produced deep images of a small sample of bright SMGs in the North Ecliptic Pole (NEP) field extracted from the NEPSC2 survey. For all 12 SMGs, we find radio-emitting mid-infrared galaxies at distances from a few arcseconds down to sub-arcsecond scales from the SMG and/or the presence of a radio-emitting AGN. The SFRs derived from the radio emission of the SMG, disentangled from the AGN-related radio emission, are systematically lower by a factor of $\sim 5$ (median value) than those derived from the multi-band SED fitting. We discuss whether our assumptions might be, at least in part, responsible for the observed discrepancy. We argue that the radio-derived SFRs are not systematically underestimated but can be affected by a significant dispersion ($0.3-0.5$ dex). Considering these new SFR estimates, the offset of the specific SFR of the 12 bright SMGs from the star-forming galaxy main sequence ($\Delta\mathrm{(SSFR)}$) is significantly reduced, with all 12 bright SMGs which are only a factor of 2 more star-forming than the main sequence galaxies. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:22:04 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Bondi",
"M.",
""
],
[
"Prandoni",
"I.",
""
],
[
"Magliocchetti",
"M.",
""
],
[
"Bisigello",
"L.",
""
],
[
"Bonato",
"M.",
""
],
[
"Giulietti",
"M.",
""
],
[
"Scaramella",
"R.",
""
],
[
"Brunetti",
"G.",
""
],
[
"Vitello",
"F.",
""
]
] |
2505.22286 | Haiquan Lu | Haiquan Lu, Yong Zeng, Shaodan Ma, Bin Li, Shi Jin, Rui Zhang | Wireless Communication for Low-Altitude Economy with UAV Swarm Enabled Two-Level Movable Antenna System | 13 pages, 10 figures | null | null | null | cs.IT eess.SP math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Unmanned aerial vehicle (UAV) is regarded as a key enabling platform for low-altitude economy, due to its advantages such as 3D maneuverability, flexible deployment, and LoS air-to-air/ground communication links. In particular, the intrinsic high mobility renders UAV especially suitable for operating as a movable antenna (MA) from the sky. In this paper, by exploiting the flexible mobility of UAV swarm and antenna position adjustment of MA, we propose a novel UAV swarm enabled two-level MA system, where UAVs not only individually deploy a local MA array, but also form a larger-scale MA system with their individual MA arrays via swarm coordination. We formulate a general optimization problem to maximize the minimum achievable rate over all ground UEs, by jointly optimizing the 3D UAV swarm placement positions, their individual MAs' positions, and receive beamforming for different UEs. We first consider the special case where each UAV has only one antenna, under different scenarios of one single UE, two UEs, and arbitrary number of UEs. In particular, for the two-UE case, we derive the optimal UAV swarm placement positions in closed-form that achieves IUI-free communication, where the UAV swarm forms a uniform sparse array (USA) satisfying collision avoidance constraint. While for the general case with arbitrary number of UEs, we propose an efficient alternating optimization algorithm to solve the formulated non-convex optimization problem. Then, we extend the results to the case where each UAV is equipped with multiple antennas. Numerical results verify that the proposed low-altitude UAV swarm enabled MA system significantly outperforms various benchmark schemes, thanks to the exploitation of two-level mobility to create more favorable channel conditions for multi-UE communications. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:22:09 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Lu",
"Haiquan",
""
],
[
"Zeng",
"Yong",
""
],
[
"Ma",
"Shaodan",
""
],
[
"Li",
"Bin",
""
],
[
"Jin",
"Shi",
""
],
[
"Zhang",
"Rui",
""
]
] |
2505.22287 | Danish Contractor | Daniel McDuff, Tim Korjakow, Kevin Klyman, Danish Contractor | New Tools are Needed for Tracking Adherence to AI Model Behavioral Use Clauses | Preprint | null | null | null | cs.CY cs.AI | http://creativecommons.org/licenses/by-sa/4.0/ | Foundation models have had a transformative impact on AI. A combination of large investments in research and development, growing sources of digital data for training, and architectures that scale with data and compute has led to models with powerful capabilities. Releasing assets is fundamental to scientific advancement and commercial enterprise. However, concerns over negligent or malicious uses of AI have led to the design of mechanisms to limit the risks of the technology. The result has been a proliferation of licenses with behavioral-use clauses and acceptable-use-policies that are increasingly being adopted by commonly used families of models (Llama, Gemma, Deepseek) and a myriad of smaller projects. We created and deployed a custom AI licenses generator to facilitate license creation and have quantitatively and qualitatively analyzed over 300 customized licenses created with this tool. Alongside this we analyzed 1.7 million models licenses on the HuggingFace model hub. Our results show increasing adoption of these licenses, interest in tools that support their creation and a convergence on common clause configurations. In this paper we take the position that tools for tracking adoption of, and adherence to, these licenses is the natural next step and urgently needed in order to ensure they have the desired impact of ensuring responsible use. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:26:55 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"McDuff",
"Daniel",
""
],
[
"Korjakow",
"Tim",
""
],
[
"Klyman",
"Kevin",
""
],
[
"Contractor",
"Danish",
""
]
] |
2505.22288 | Jan Speller | Jan Speller, Malte Luttermann, Marcel Gehrke, Tanya Braun | Compression versus Accuracy: A Hierarchy of Lifted Models | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Probabilistic graphical models that encode indistinguishable objects and relations among them use first-order logic constructs to compress a propositional factorised model for more efficient (lifted) inference. To obtain a lifted representation, the state-of-the-art algorithm Advanced Colour Passing (ACP) groups factors that represent matching distributions. In an approximate version using $\varepsilon$ as a hyperparameter, factors are grouped that differ by a factor of at most $(1\pm \varepsilon)$. However, finding a suitable $\varepsilon$ is not obvious and may need a lot of exploration, possibly requiring many ACP runs with different $\varepsilon$ values. Additionally, varying $\varepsilon$ can yield wildly different models, leading to decreased interpretability. Therefore, this paper presents a hierarchical approach to lifted model construction that is hyperparameter-free. It efficiently computes a hierarchy of $\varepsilon$ values that ensures a hierarchy of models, meaning that once factors are grouped together given some $\varepsilon$, these factors will be grouped together for larger $\varepsilon$ as well. The hierarchy of $\varepsilon$ values also leads to a hierarchy of error bounds. This allows for explicitly weighing compression versus accuracy when choosing specific $\varepsilon$ values to run ACP with and enables interpretability between the different models. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:27:32 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Speller",
"Jan",
""
],
[
"Luttermann",
"Malte",
""
],
[
"Gehrke",
"Marcel",
""
],
[
"Braun",
"Tanya",
""
]
] |
2505.22289 | Yan Zhang | Yan Zhang, Jun Liao, Xinyan Fan, Kuangnan Fang, Yuhong Yang | Network Model Averaging Prediction for Latent Space Models by K-Fold Edge Cross-Validation | null | null | null | null | stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In complex systems, networks represent connectivity relationships between nodes through edges. Latent space models are crucial in analyzing network data for tasks like community detection and link prediction due to their interpretability and visualization capabilities. However, when the network size is relatively small, and the true latent space dimension is considerable, the parameters in latent space models may not be estimated very well. To address this issue, we propose a Network Model Averaging (NetMA) method tailored for latent space models with varying dimensions, specifically focusing on link prediction in networks. For both single-layer and multi-layer networks, we first establish the asymptotic optimality of the proposed averaging prediction in the sense of achieving the lowest possible prediction loss. Then we show that when the candidate models contain some correct models, our method assigns all weights to the correct models. Furthermore, we demonstrate the consistency of the NetMA-based weight estimator tending to the optimal weight vector. Extensive simulation studies show that NetMA performs better than simple averaging and model selection methods, and even outperforms the "oracle" method when the real latent space dimension is relatively large. Evaluation on collaboration and virtual event networks further emphasizes the competitiveness of NetMA in link prediction performance. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:27:39 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Zhang",
"Yan",
""
],
[
"Liao",
"Jun",
""
],
[
"Fan",
"Xinyan",
""
],
[
"Fang",
"Kuangnan",
""
],
[
"Yang",
"Yuhong",
""
]
] |
2505.22290 | Fanzeng Xia | Fanzeng Xia, Yidong Luo, Tinko Sebastian Bartels, Yaqi Xu, and Tongxin Li | Rethinking the Unsolvable: When In-Context Search Meets Test-Time Scaling | null | null | null | null | cs.AI cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Recent research has highlighted that Large Language Models (LLMs), even when trained to generate extended long reasoning steps, still face significant challenges on hard reasoning problems. However, much of the existing literature relies on direct prompting with simple in-context learning examples for evaluation, which largely overlooks advanced techniques to elicit LLMs' deliberate reasoning before drawing conclusions that LLMs hit a performance ceiling. In this paper, we systematically explore the combined potential of in-context search and test-time scaling on super hard reasoning tasks. We find that by employing advanced in-context search prompting to LLMs augmented with internal scaling, one can achieve transformative performance breakthroughs on tasks previously deemed "unsolvable" (e.g., reported success rates below 5%). We provide both empirical results and theoretical analysis of how this combination can unleash LLM reasoning capabilities: i) Empirically, on controlled NP-hard tasks and complex real-world planning benchmarks, our approach achieves up to a 30x improvement in success rates compared to previously reported results without any external mechanisms; ii) Theoretically, we show that in-context search prompting, when combined with internal scaling, significantly extends the complexity class of solvable reasoning problems. These findings challenge prevailing assumptions about the limitations of LLMs on complex tasks, indicating that current evaluation paradigms systematically underestimate their true potential. Our work calls for a critical reassessment of how LLM reasoning is benchmarked and a more robust evaluation strategy that fully captures the true capabilities of contemporary LLMs, which can lead to a better understanding of their operational reasoning boundaries in real-world deployments. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:28:18 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Xia",
"Fanzeng",
""
],
[
"Luo",
"Yidong",
""
],
[
"Bartels",
"Tinko Sebastian",
""
],
[
"Xu",
"Yaqi",
""
],
[
"Li",
"Tongxin",
""
]
] |
2505.22291 | Saptarshi Neil Sinha | Saptarshi Neil Sinha and P. Julius Kuehn and Johannes Koppe and Arjan Kuijper and Michael Weinmann | Neural Restoration of Greening Defects in Historical Autochrome Photographs Based on Purely Synthetic Data | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | The preservation of early visual arts, particularly color photographs, is challenged by deterioration caused by aging and improper storage, leading to issues like blurring, scratches, color bleeding, and fading defects. In this paper, we present the first approach for the automatic removal of greening color defects in digitized autochrome photographs. Our main contributions include a method based on synthetic dataset generation and the use of generative AI with a carefully designed loss function for the restoration of visual arts. To address the lack of suitable training datasets for analyzing greening defects in damaged autochromes, we introduce a novel approach for accurately simulating such defects in synthetic data. We also propose a modified weighted loss function for the ChaIR method to account for color imbalances between defected and non-defected areas. While existing methods struggle with accurately reproducing original colors and may require significant manual effort, our method allows for efficient restoration with reduced time requirements. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:28:35 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Sinha",
"Saptarshi Neil",
""
],
[
"Kuehn",
"P. Julius",
""
],
[
"Koppe",
"Johannes",
""
],
[
"Kuijper",
"Arjan",
""
],
[
"Weinmann",
"Michael",
""
]
] |
2505.22292 | Ernesto Contreras | E. I\~niguez, M. Freire, L. Leal and E. Contreras | A topological invariant in the context of the loop representation of the massive Kalb-Ramond-Klein-Gordon model | null | Phys. Lett. A 552, 130672 (2025) | 10.1016/j.physleta.2025.130672 | null | hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We employ the Dirac procedure to quantize the self-dual massive Kalb-Ramond-Klein-Gordon model in $2+1$ dimensional spacetimes. The canonical fields are expressed in terms of $2$-surfaces and signed points, ensuring the automatic realization of the quantum algebra. As the duality rotation preserving the action can be implemented infinitesimally, we derive the conserved quantity that generates the transformation. Given that such a generator is a two dimensional topological quantity, its representation in terms of geometrical operators yields a two dimensional invariant (reminiscent of a projection of Gauss's law in electrodynamics), which encodes the same information of the well-known winding number. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:28:53 GMT"
}
] | 2025-05-29T00:00:00 | [
[
"Iñiguez",
"E.",
""
],
[
"Freire",
"M.",
""
],
[
"Leal",
"L.",
""
],
[
"Contreras",
"E.",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.