May 2025 arXiv papers — page 95

Dheeraj Baby, Yifei Tang, Hieu Duy Nguyen, Yu-Xiang Wang

In this paper, we study the problem of estimation and learning under temporal distribution shift. Consider an observation sequence of length $n$, which is a noisy realization of a time-varying groundtruth sequence. Our focus is to develop methods to estimate the groundtruth at the final time-step while providing sharp point-wise estimation error rates. We sh

Sarah E. Wessinger, Leslie N. Smith, Jacob Gull, Jonathan Gehman

Accurately estimating the refractive environment over multiple frequencies within the marine atmospheric boundary layer is crucial for the effective deployment of radar technologies. Traditional parabolic equation simulations, while effective, can be computationally expensive and time-intensive, limiting their practical application. This communication explor

Yue Fan, Xuehai He, Diji Yang, Kaizhi Zheng

Recent studies have demonstrated the efficacy of using Reinforcement Learning (RL) in building reasoning models that articulate chains of thoughts prior to producing final answers. However, despite ongoing advances that aim at enabling reasoning for vision-language tasks, existing open-source visual reasoning models typically generate reasoning content with

Domonkos Svastits, Bence Hetényi, Gábor Széchenyi, James Wootton

Qubit readout schemes often deviate from ideal projective measurements, introducing critical issues that limit quantum computing performance. In this work, we model charge-sensing-based readout for semiconductor spin qubits in double quantum dots, and identify key error mechanisms caused by the back-action of the charge sensor. We quantify how the charge noi

Yuchen Yan, Jin Jiang, Zhenbang Ren, Yijun Li

Large reasoning models such as OpenAI o1 and DeepSeek-R1 have demonstrated remarkable performance in complex reasoning tasks. A critical component of their training is the incorporation of reference-based reward systems within reinforcement learning (RL), where model outputs are evaluated against ground truth references. However, existing reward benchmarks f

Ruizhi Shao, Yinghao Xu, Yujun Shen, Ceyuan Yang

Generating photorealistic videos of digital humans in a controllable manner is crucial for a plethora of applications. Existing approaches either build on methods that employ template-based 3D representations or emerging video generation models but suffer from poor quality or limited consistency and identity preservation when generating individual or multipl

David M. Rothschild, Markus Mobius, Jake M. Hofman, Eleanor W. Dillon

Generative AI has transformed human-computer interaction by enabling natural language interfaces and the emergence of autonomous agents capable of acting on users' behalf. While early applications have improved individual productivity, these gains have largely been confined to predefined tasks within existing workflows. We argue that the more profound econom

Taehoon Kim, Henry Gouk, Minyoung Kim, Timothy Hospedales

Certifying the IID generalisation ability of deep networks is the first of many requirements for trusting AI in high-stakes applications from medicine to security. However, when instantiating generalisation bounds for deep networks it remains challenging to obtain non-vacuous guarantees, especially when applying contemporary large models on the small scale d

Ivan Homoliak, Tomáš Švondr

Voting is a cornerstone of democracy, allowing citizens to express their will and make collective decisions. With advancing technology, online voting is gaining popularity as it enables voting from anywhere with Internet access, eliminating the need for printed ballots or polling stations. However, despite its benefits, online voting carries significant risk

Abdalrhman Mohamed, Tomaz Mascarenhas, Harun Khan, Haniel Barbosa

Lean is an increasingly popular proof assistant based on dependent type theory. Despite its success, it still lacks important automation features present in more seasoned proof assistants, such as the Sledgehammer tactic in Isabelle/HOL. A key aspect of Sledgehammer is the use of proof-producing SMT solvers to prove a translated proof goal and the reconstruc

Lisa Alazraki, Tan Yi-Chern, Jon Ander Campos, Maximilian Mozes

The capabilities of Large Language Models (LLMs) are routinely evaluated by other LLMs trained to predict human preferences. This framework--known as LLM-as-a-judge--is highly scalable and relatively low cost. However, it is also vulnerable to malicious exploitation, as LLM responses can be tuned to overfit the preferences of the judge. Previous work shows t

Daniel Hlubinka, Šárka Hudecová

A multivariate one-sample location test based on the center-outward ranks and signs is considered, and two different testing procedures are proposed for centrally symmetric distributions. The first test is based on a random division of the data into two samples, while the second one uses a symmetrized sample. The asymptotic distributions of the proposed test

Zhiwen Chen, Bo Leng, Zhuoren Li, Hanming Deng

Integrating Large Language Models (LLMs) with Reinforcement Learning (RL) can enhance autonomous driving (AD) performance in complex scenarios. However, current LLM-Dominated RL methods over-rely on LLM outputs, which are prone to hallucinations. Evaluations show that state-of-the-art LLM indicates a non-hallucination rate of only approximately 57.95% when a

Danna Zheng, Mirella Lapata, Jeff Z. Pan

Information alignment evaluators are vital for various NLG evaluation tasks and trustworthy LLM deployment, reducing hallucinations and enhancing user trust. Current fine-grained methods, like FactScore, verify facts individually but neglect inter-fact dependencies, enabling subtle vulnerabilities. In this work, we introduce MontageLie, a challenging benchma

Fengyuan Dai, Zifeng Zhuang, Yufei Huang, Siteng Huang

Diffusion models have emerged as powerful generative tools across various domains, yet tailoring pre-trained models to exhibit specific desirable properties remains challenging. While reinforcement learning (RL) offers a promising solution,current methods struggle to simultaneously achieve stable, efficient fine-tuning and support non-differentiable rewards.

Minjung Park, Jodi Forlizzi, John Zimmerman

Innovators transform the world by understanding where services are successfully meeting customers' needs and then using this knowledge to identify failsafe opportunities for innovation. Pre-trained models have changed the AI innovation landscape, making it faster and easier to create new AI products and services. Understanding where pre-trained models are su

Alessandro Greco, Zachary Slepian, Jiamin Hou, Alex Krolewski

We show that the Cosmic Microwave Background (CMB) lensing trispectrum is sensitive to parity violation in Large-Scale Structure (LSS). We obtain a compact expression for the reduced lensing trispectrum that is general for any input matter trispectrum. We then present as an example a simple parity-violating toy model for the latter, and explicitly compute th

Felix Schmidt, Alexander Roth, Wolf-Peter Schill

Hydrogen-based long-duration electricity storage (LDES) is a key component of renewable energy systems to deal with seasonality and prolonged periods of low wind and solar energy availability. In this paper, we investigate how electrified heating with heat pumps impacts LDES requirements in a fully renewable European energy system, and which role thermal sto

Javier M. Lizana

In these proceedings, I present a composite Higgs model in which the flavour hierarchies between the third and light families emerge naturally. In particular, CKM mixing angles turn out to be suppressed while PMNS matrix remains anarchic. This flavour structure arises as a consequence of the extended non-universal gauge symmetry of the model and the electrow

Tobias Barthel

A theorem of Cohen from 1950 states that a commutative ring is Noetherian if and only if every prime ideal is finitely generated. In this note, we establish analogues of this result in tensor triangular geometry. In particular, for an essentially small tensor triangulated category $\mathscr{K}$ with weakly Noetherian spectrum, we show that every prime ideal

Siting Li, Xiang Gao, Simon Shaolei Du

While an image is worth more than a thousand words, only a few provide crucial information for a given task and thus should be focused on. In light of this, ideal text-to-image (T2I) retrievers should prioritize specific visual attributes relevant to queries. To evaluate current retrievers on handling attribute-focused queries, we build COCO-Facet, a COCO-ba

Stefano Morisi, Giovanna Paola Perdonà, Giulia Ricciardi

Among the particles being searched for at the LHC beyond the Standard Model are bileptons, which are doubly charged gauge bosons. Bileptons are predicted by several Standard Model extensions, including the so-called 331 models with $\beta = \sqrt 3$. The minimal formulation of these models is generally plagued by a "low energy" Landau pole, which can undermi

Jun Wan, Lingrui Mei

The rapid advancement of large language models (LLMs) calls for a rigorous theoretical framework to explain their empirical success. While significant progress has been made in understanding LLM behaviors, existing theoretical frameworks remain fragmented in explaining emergent phenomena through a unified mathematical lens. We establish the first formal conn

Reza Gheissari, Allan Sly, Youngtak Sohn

We consider the Ising, and more generally, $q$-state Potts Glauber dynamics on random $d$-regular graphs on $n$ vertices at low temperatures $\beta \gtrsim \frac{\log d}{d}$. The mixing time is exponential in $n$ due to a bottleneck between $q$ dominant phases consisting of configurations in which the majority of vertices are in the same state. We prove that

Pedro P. Santos, Alberto Sardinha, Francisco S. Melo

In this work, we contribute the first approach to solve infinite-horizon discounted general-utility Markov decision processes (GUMDPs) in the single-trial regime, i.e., when the agent's performance is evaluated based on a single trajectory. First, we provide some fundamental results regarding policy optimization in the single-trial regime, investigating whic

Xinyin Ma, Runpeng Yu, Gongfan Fang, Xinchao Wang

Diffusion Language Models (DLMs) have been seen as a promising competitor for autoregressive language models. However, diffusion language models have long been constrained by slow inference. A core challenge is that their non-autoregressive architecture and bidirectional attention preclude the key-value cache that accelerates decoding. We address this bottle

Debraj Chakrabarti, Prachi Mahajan

We describe a method of defining a Hermitian metric on Kobayashi hyperbolic manifolds. The metric is distance decreasing under holomorphic mappings, up to a multiplicative constant. This method is distinct from the classical construction of Wu, and yields a metric which is expected to have superior regularity properties.

Chuanhao Li, Jianwen Sun, Yukang Feng, Mingliang Zhai

Current text-to-image (T2I) generation models achieve promising results, but they fail on the scenarios where the knowledge implied in the text prompt is uncertain. For example, a T2I model released in February would struggle to generate a suitable poster for a movie premiering in April, because the character designs and styles are uncertain to the model. To

Zhen Zhang, Xuehai He, Weixiang Yan, Ao Shen

Human cognition typically involves thinking through abstract, fluid concepts rather than strictly using discrete linguistic tokens. Current reasoning models, however, are constrained to reasoning within the boundaries of human language, processing discrete token embeddings that represent fixed points in the semantic space. This discrete constraint restricts

Jorge Bacca

Deep learning-based models have demonstrated remarkable success in solving illposed inverse problems; however, many fail to strictly adhere to the physical constraints imposed by the measurement process. In this work, we introduce a projection-based correction method to enhance the inference of deep inverse networks by ensuring consistency with the forward m

Changtai Zhu, Siyin Wang, Ruijun Feng, Kai Song

Conversational search systems require effective handling of context-dependent queries that often contain ambiguity, omission, and coreference. Conversational Query Reformulation (CQR) addresses this challenge by transforming these queries into self-contained forms suitable for off-the-shelf retrievers. However, existing CQR approaches suffer from two critica

Lijun Ding, Haihao Lu, Jinwen Yang

Augmented Lagrangian Method (ALM) combined with Burer-Monteiro (BM) factorization, dubbed ALM-BM, offers a powerful approach for solving large-scale low-rank semidefinite programs (SDPs). Despite its empirical success, the theoretical understandings of the resulting non-convex ALM-BM subproblems, particularly concerning their structural properties and effici

Huanxuan Liao, Wen Hu, Yao Xu, Shizhu He

Large Language Models (LLMs) encounter significant challenges in long-sequence inference due to computational inefficiency and redundant processing, driving interest in context compression techniques. Existing methods often rely on token importance to perform hard local compression or encode context into latent representations for soft global compression. Ho

Yu-Xiang Luo, Yi-Cheng Lin, Ming-To Chuang, Jia-Hung Chen

Despite extensive research on toxic speech detection in text, a critical gap remains in handling spoken Mandarin audio. The lack of annotated datasets that capture the unique prosodic cues and culturally specific expressions in Mandarin leaves spoken toxicity underexplored. To address this, we introduce ToxicTone -- the largest public dataset of its kind --

Yifan Cheng, Ruoyi Zhang, Jiatong Shi

Acquiring large-scale emotional speech data with strong consistency remains a challenge for speech synthesis. This paper presents MIKU-PAL, a fully automated multimodal pipeline for extracting high-consistency emotional speech from unlabeled video data. Leveraging face detection and tracking algorithms, we developed an automatic emotion analysis system using

Romain Mottier, Alexandre Ern, Laurent Guillot

Hybrid high-order (HHO) methods are numerical methods characterized by several interesting properties such as local conservativity, geometric flexibility and high-order accuracy. Here, HHO schemes are studied for the space semi-discretization of coupled elasto-acoustic waves in the time domain using a first-order formulation. Explicit and singly diagonal imp

Mikhail Budnikov, Ivan Yamshchikov

This work explores transfer learning from several synthetic languages to English. We investigate the structure of the embeddings in the fine-tuned models, the information they contain, and the capabilities of the fine-tuned models on simple linguistic tasks. We also introduce a new synthetic language that leads to better transfer to English than the language

Jaiane Santos, Carlos Bengaly, Rodrigo S. Gonçalves

One of the most crucial tests of the standard cosmological model consists on testing possible variations on fundamental physical constants. In frameworks such as the minimally extended varying speed of light model (meVSL), the relationship between the luminosity distance ($D_{\text{L}}$) and the angular diameter distance ($D_{\text{A}}$), namely the cosmic d

Sergei Balashev, Pasquier Noterdaeme, Neeraj Gupta, Jens-Kristian Krogager

Quasars, powered by gas accretion onto supermassive black holes, rank among the most energetic objects of the Universe. While they are thought to be ignited by galaxy mergers and affect the surrounding gas, observational constraints on both processes remain scarce. Here we unveil a major merging system at redshift $z \approx 2.7$, and demonstrate that radiat

Taimoor Ahmad

The integration of unmanned aerial vehicles (UAVs) into smart agriculture has enabled real-time monitoring, data collection, and automated farming operations. However, the high mobility, decentralized nature, and low-power communication of UAVs pose significant security challenges, particularly in ensuring transaction integrity and trust. This paper presents

Kaizhi Zheng, Ruijian Zha, Zishuo Xu, Jing Gu

Acquiring detailed 3D scenes typically demands costly equipment, multi-view data, or labor-intensive modeling. Therefore, a lightweight alternative, generating complex 3D scenes from a single top-down image, plays an essential role in real-world applications. While recent 3D generative models have achieved remarkable results at the object level, their extens

Wen Yin

Paleo detectors are emerging dark matter detection technology that exploits ancient minerals as passive, time-integrated detectors. Unlike conventional real-time experiments, they search for permanent damage tracks-typically tens of nanometers to micrometers long-left in crystal lattices by rare particle interactions, most notably dark matter induced nuclear

Bo Hu, Joon Y. Park, Junhui Qian

This paper introduces a novel approach to investigate the dynamics of state distributions, which accommodate both cross-sectional distributions of repeated panels and intra-period distributions of a time series observed at high frequency. In our approach, densities of the state distributions are regarded as functional elements in a Hilbert space, and are ass

Chirudeep Tupakula, Rittika Shamsuddin

Machine learning systems are increasingly deployed in high-stakes domains, yet they remain vulnerable to bias systematic disparities that disproportionately impact specific demographic groups. Traditional bias detection methods often depend on access to sensitive labels or rely on rigid fairness metrics, limiting their applicability in real-world settings. T

Pushpa Pandey, Hao Lu, James Maxwell, James Brock

We report high nuclear polarization of 1.2 mbar 3He gas in a sealed cell in magnetic fields up to 5 T using Metastability Exchange Optical Pumping (MEOP). The creation of a highly polarized 3He gas target for use in the 5 T field of Jefferson Lab's CLAS12 spectrometer would enable new studies of spin-dependent asymmetries on the neutron. A systematic study w

Michael I. Ganzburg

We prove $L_q(\R^m)$--discretization inequalities for entire functions $f$ of exponential type in the form \ba C_2\|f\|_{L_q(\R^m)} \le \left(\sum_{\nu=1}^\iy \left\vert f\left(X_\nu\right) \right\vert^q\right)^{1/q} \le C_1\|f\|_{L_q(\R^m)},\qquad q\in[1,\iy], \ea with estimates for $C_1$ and $C_2$. We find a necessary and sufficient condition on $\Omega=\l

Nansen Petrosyan

We exhibit finitely generated torsion-free groups for which any action on any finite-dimensional CW-complex with finite Betti numbers has a global fixed point.

Shenghe Zheng, Qianjia Cheng, Junchi Yao, Mengsong Wu

Large Language Models (LLMs) have achieved remarkable progress on advanced reasoning tasks such as mathematics and coding competitions. Meanwhile, physics, despite being both reasoning-intensive and essential to real-world understanding, received limited academic and industrial attention. This paper introduces PHYSICS, a dataset containing 16,568 high-qualit

Taimoor Ahmad

The growing complexity of cyber threats has rendered static firewalls increasingly ineffective for dynamic, real-time intrusion prevention. This paper proposes a novel AI-driven dynamic firewall optimization framework that leverages deep reinforcement learning (DRL) to autonomously adapt and update firewall rules in response to evolving network threats. Our

Pablo Rocha

Let $\mathbb{H}^{n}$ be the Heisenberg group and $Q = 2n+2$. For $1 < q < \infty$, $\gamma > 0$ and an exponent function $p(\cdot)$ on $\mathbb{H}^n$, which satisfy log-H\"older conditions, with $0 < p_{-} \leq p_{+} < \infty$, we introduce the variable Calder\'on-Hardy spaces $\mathcal{H}^{p(\cdot)}_{q, \gamma}(\mathbb{H}^{n})$, and show for every $f \in H^

Taimoor Ahmad

Vehicular Fog Computing (VFC) is a promising paradigm to meet the low-latency and high-bandwidth demands of Intelligent Transportation Systems (ITS). However, dynamic vehicle mobility and diverse trust boundaries introduce critical security challenges. This paper presents a novel Zero-Trust Mobility-Aware Authentication Framework (ZTMAF) for secure communica

Tianyi Pan, Hwashin Hyun Shin, Glen McGee, Alex Stringer

Quantifying associations between short-term exposure to ambient air pollution and health outcomes is an important public health priority. Many studies have investigated the association considering delayed effects within the past few days. Adaptive cumulative exposure distributed lag non-linear models (ACE-DLNMs) quantify associations between health outcomes

Dániel Hajtó, Waleed El-Geresy, Deniz Gündüz, György Cserey

Knowing how to reliably use memristors as information storage devices is crucial not only to their role as emerging memories, but also for their application in neural network acceleration and as components of novel neuromorphic systems. In order to better understand the dynamics of information storage on memristors, it is essential to be able to characterise

Shoya Kise, Takesa Uehara, Takashi Shinzato

Recent interest in noncircular trigonometric proofs has underscored the need for alternative methodologies. Jackson and Johnson's 2024 study addresses a longstanding gap in the foundations of trigonometric proofs. Inspired by the work of Jackson and Johnson [JJ24], we present three noncircular proofs of the Pythagorean theorem based on trigonometric identiti

Weihao Xia, Cengiz Oztireli

The intrication of brain signals drives research that leverages multimodal AI to align brain modalities with visual and textual data for explainable descriptions. However, most existing studies are limited to coarse interpretations, lacking essential details on object descriptions, locations, attributes, and their relationships. This leads to imprecise and a

Palash Chatterjee, Roni Khardon

Continuous time systems are often modeled using discrete time dynamics but this requires a small simulation step to maintain accuracy. In turn, this requires a large planning horizon which leads to computationally demanding planning problems and reduced performance. Previous work in model-free reinforcement learning has partially addressed this issue using a

Taiye Chen, Zeming Wei, Ang Li, Yisen Wang

Large Language Models (LLMs) are known to be vulnerable to jailbreaking attacks, wherein adversaries exploit carefully engineered prompts to induce harmful or unethical responses. Such threats have raised critical concerns about the safety and reliability of LLMs in real-world deployment. While existing defense mechanisms partially mitigate such risks, subse

Swarnabha Chattaraj, Giulia Galli

Radiative and nonradiative resonant couplings between defects are ubiquitous phenomena in photonic devices used in classical and quantum information technology applications. In this work we present a first principles approach to enable quantitative predictions of the energy transfer between defects in photonic cavities, beyond the dipole-dipole approximation

Jack Kelly, Devarshi Mukherjee

We study several categories of analytic stacks relative to the category of bornological modules over a Banach ring. When the underlying Banach ring is a non-Archimedean valued field, this category contains derived rigid analytic spaces as a full subcategory. When the underlying field is the complex numbers, it contains the category of derived complex analyti

Pritthijit Biswas, Jaya NN Iyer

Let $X$ be a smooth projective curve of genus $g$ over the field $\mathbb{C}$. Let $M_{X}(2,L)$ denote the moduli space of stable rank $2$ vector bundles on $X$ with fixed determinant $L$ of degree $2g-1$. Consider the Brill-Noether subvariety $W^{1}_{X}(2,L)$ of $M_{X}(2,L)$ which parametrises stable vector bundles having at least two linearly independent g

Sandeep Kumar Verma, Athulya P

In this work, we introduce the $\beta$-semigroup for $\beta > 0$, which unifies and extends the classical Poisson (for $\beta=1$) and heat (for $\beta=2$) semigroups within the Dunkl analysis framework. Leveraging this semigroup, we derive an explicit representation for the inverse of the Dunkl-Riesz potential and characterize the image of the function space

Kanan Kiguchi, Yunhao Tu, Katsuhiro Ajito, Fady Alnajjar

We propose a novel framework for integrating fragmented multi-modal data in Alzheimer's disease (AD) research using large language models (LLMs) and knowledge graphs. While traditional multimodal analysis requires matched patient IDs across datasets, our approach demonstrates population-level integration of MRI, gene expression, biomarkers, EEG, and clinical

Jingzhe Liu, Zhigang Hua, Yan Xie, Bingheng Li

Temporal Graph Neural Networks (TGNNs) have gained growing attention for modeling and predicting structures in temporal graphs. However, existing TGNNs primarily focus on pairwise interactions while overlooking higher-order structures that are integral to link formation and evolution in real-world temporal graphs. Meanwhile, these models often suffer from ef

Prashant Gupta, Jasmin Bedow, Eric Mascot, Dirk K. Morr

We propose that the existence of vortices in FeSe$_{1-x}$Te$_x$ with and without Majoarana zero modes (MZMs) can be explained by a heterogeneous mixture of strong topological and trivial superconducting domains, with only vortices in the former exhibiting MZMs. We identify the spectroscopic signatures of topological and trivial vortices and show that they ar

Dipendra Prasad

The Weak approximation theorem describes the closure of $G(Q)$ inside $G(Q_p)$ as well as inside $G(R)$ for $G$ an algebraic group over $Q$; the closure is always an open normal subgroup with finite abelian quotient, and is well understood in a certain sense even if precise results are not always available (such as for tori!). In this paper, for a finitely g

Darja Smite, Nils Brede Moe, Maria Teresa Baldassarre, Fabio Calefato

The COVID-19 pandemic has permanently altered workplace structures, making remote work a widespread practice. While many employees advocate for flexibility, many employers reconsider their attitude toward remote work and opt for structured return-to-office mandates. Media headlines repeatedly emphasize that the corporate world is returning to full-time offic

Adam Gould, Francesca Toni

We introduce Gradual Abstract Argumentation for Case-Based Reasoning (Gradual AA-CBR), a data-driven, neurosymbolic classification model in which the outcome is determined by an argumentation debate structure that is learned simultaneously with neural-based feature extractors. Each argument in the debate is an observed case from the training data, favouring

Dikshit Chauhan, Bapi Dutta, Indu Bala, Niki van Stein

Large Language Models (LLMs) and Evolutionary Computation (EC) are increasingly being combined to support automated optimization, algorithm design, and adaptive decision-making. This survey reviews the bidirectional interaction between these two paradigms and examines how their complementary strengths can be leveraged in hybrid intelligent systems. First, we

Weiming Wu, Jin Ye, Zi-kang Wang, Zhi Zhou

Obtaining large-scale, high-quality reasoning data is crucial for improving the geometric reasoning capabilities of multi-modal large language models (MLLMs). However, existing data generation methods, whether based on predefined tem plates or constrained symbolic provers, inevitably face diversity and numerical generalization limitations. To address these l

Jilin Hu, Jianyu Zhang, Yongwang Zhao, Talia Ringer

Formal methods play a crucial role in ensuring the reliability of critical systems through rigorous mathematical verification. However, their adoption remains limited due to the labor-intensive nature of manual proof construction. Recent advances in large language models (LLMs) have opened new opportunities for automated theorem proving. Two main paradigms h

Xiaoxue Yang, Bozhidar Stevanoski, Matthieu Meeus, Yves-Alexandre de Montjoye

Large language models (LLMs) are increasingly deployed in real-world applications ranging from chatbots to agentic systems, where they are expected to process untrusted data and follow trusted instructions. Failure to distinguish between the two poses significant security risks, exploited by prompt injection attacks, which inject malicious instructions into

Zhuodong Jiang, Haoran Wang, Guoxi Huang, Brett Seymour

Reconstructing high-fidelity underwater scenes remains a challenging task due to light absorption, scattering, and limited visibility inherent in aquatic environments. This paper presents an enhanced Gaussian Splatting-based framework that improves both the visual quality and geometric accuracy of deep underwater rendering. We propose decoupled learning for

Piyali Ganguly, Priyanka Rani, Gulab C. Dewangan

We present high-resolution near-ultraviolet (NUV) and far-ultraviolet (FUV) deep imaging of the field around the Seyfert galaxy IC~4329A based on five observations performed with the Ultra-Violet Imaging Telescope (UVIT), onboard AstroSat. The long exposures of 82.9 ks in NUV (N245M; $\lambda_{mean}=2447$\r{A} ; $\Delta\lambda = 270$\r{A}) and 92.2~ks in FUV

Yanara Marks, Jessie Cunningham, Arlene Jiang, Linke Li

Utilizing Bayesian methods in clinical trials has become increasingly popular, as they can incorporate historical data and expert opinions into the design and allow for smaller sample sizes to reduce costs while providing reliable and robust statistical results. Sample size determination (SSD) is a key aspect of clinical trial design and various methods for

Gaurav Srivastava, Zhenyu Bi, Meng Lu, Xuan Wang

Large language models (LLMs) have improved significantly in their reasoning through extensive training on massive datasets. However, relying solely on additional data for improvement is becoming increasingly impractical, highlighting the need for models to autonomously enhance their reasoning without external supervision. In this paper, we propose Debate, Tr

Yuchen Dong, Zhengsong Lu, Xiaoyu Cao, Zhengwen He

This paper presents a distributionally robust planning method for hydrogen-electrical microgrids over islands, where the cross-island energy exchange is supported by a maritime hydrogen transport network. This planning problem is complicated due to heterogeneous off-shore wind-driven uncertainties (i.e., renewable power, transport availability, demand fluctu

Wayne Yu, Trevor Williams, Russell Carpenter

The Fermi Gamma ray Space Telescope, launched in 2008, has over 16 years of operations providing gamma ray (8 keV to 300 Gev) spectra science observations of cosmic phenomena. It continues to provide invaluable research for the astrophysics community which include the study of pulsars, cosmic rays, gamma ray bursts, and coordination with gravity wave observa

Maleen Kidiwela, Marine A. Denolle, William S. D. Wilcock, Kuan-Fu Feng

Complex interactions between strain accumulation, fault slip, and fluid migration influence shallow subduction zone dynamics. Using a decade of continuous ambient seismic data from Cascadia seafloor observatories, we identified distinct regional variations in subduction dynamics. Northern Cascadia exhibits a fully locked megathrust with persistent strain acc

Sarfraz Ahmad, Momina Ahsan, Muhammad Nabeel Asim, Andreas Dengel

The identification of bitter peptides is crucial in various domains, including food science, drug discovery, and biochemical research. These peptides not only contribute to the undesirable taste of hydrolyzed proteins but also play key roles in physiological and pharmacological processes. However, experimental methods for identifying bitter peptides are time

Dillon Plunkett, Adam Morris, Keerthi Reddy, Jorge Morales

We have only limited understanding of how and why large language models (LLMs) respond in the ways that they do. Their neural networks have proven challenging to interpret, and we are only beginning to tease out the function of individual neurons and circuits within them. However, another path to understanding these systems is to investigate and develop thei

Smain Amari, Mustafa Bakr, Uwe Rosenberg

Building blocks containing strongly coupled posts offer new possibilities for advanced coaxial (comb-line) filter designs. Equivalent circuits based on the individual resonances of the posts cannot be used to reliably describe the behavior of these structures because of the strong coupling between the posts. Instead, sets of electromagnetic (EM) resonances t

Shenghe Zheng, Hongzhi Wang, Chenyu Huang, Xiaohui Wang

With more open-source models available for diverse tasks, model merging has gained attention by combining models into one, reducing training, storage, and inference costs. Current research mainly focuses on model merging for full fine-tuning, overlooking the popular LoRA. However, our empirical analysis reveals that: a) existing merging methods designed for

Luqin Gan, Tarek M. Zikry, Genevera I. Allen

As machine learning systems are increasingly used in high-stakes domains, there is a growing emphasis placed on making them interpretable to improve trust in these systems. In response, a range of interpretable machine learning (IML) methods have been developed to generate human-understandable insights into otherwise black box models. With these methods, a f

Heyang Liu, Yuhao Wang, Ziyang Cheng, Hongcheng Liu

Speech large language models (SpeechLLMs) have extended human-machine interactions from the text modality to the dynamic speech domain. Spoken dialogues convey diverse information, including semantic concepts, acoustic variations, paralanguage cues, and environmental context. However, existing evaluations of speech interaction models lack instances mimicking

Anton Nazarov, Anton Selemenchuk

Consider an $n\times k$ matrix of i.i.d. Bernoulli random numbers with $p=1/2$. Dual RSK algorithm gives a bijection of this matrix to a pair of Young tableaux of conjugate shape, which is manifestation of skew Howe $GL_{n}\times GL_{k}$-duality. Thus the probability measure on zero-ones matrix leads to the probability measure on Young diagrams proportional

Xiangyu Wang, Donglin Yang, Yue Liao, Wenhao Zheng

Unmanned Aerial Vehicles (UAVs) are evolving into language-interactive platforms, enabling more intuitive forms of human-drone interaction. While prior works have primarily focused on high-level planning and long-horizon navigation, we shift attention to language-guided fine-grained trajectory control, where UAVs execute short-range, reactive flight behavior

Lara Bohnenblust, Harald Ita, Manfred Kraus, Johannes Schlenk

We are employing a supersymmetric variant of the worldline quantum field theory (WQFT) formalism to compute the far-field momentum-space gravitational waveform emitted during the scattering of two spinning black holes at next-to-leading order (NLO) in the post-Minkowskian expansion. Our results are accurate up to quadratic-in-spin contributions, which means

Abdur Rehman, Daniel Floryan

Swimming and flying animals demonstrate remarkable adaptations to diverse flow conditions in their environments. In this study, we aim to advance the fundamental understanding of the interaction between flexible bodies and heterogeneous flow conditions. We develop a linear inviscid model of an elastically mounted foil that passively pitches in response to a

Zongru Shao, Xin Wang, Zhanyang Liu, Chenhan Wang

Recent research leverages large language models (LLMs) for early mental health detection, such as depression, often optimized with machine-generated data. However, their detection may be subject to unknown weaknesses. Meanwhile, quality control has not been applied to these generated corpora besides limited human verifications. Our goal is to systematically

Xiaoyu Luo, Yiyi Chen, Johannes Bjerva, Qiongxiu Li

We present the first comprehensive study of Memorization in Multilingual Large Language Models (MLLMs), analyzing 95 languages using models across diverse model scales, architectures, and memorization definitions. As MLLMs are increasingly deployed, understanding their memorization behavior has become critical. Yet prior work has focused primarily on monolin

Coby Penso, Bar Mahpud, Jacob Goldberger, Or Sheffet

Conformal prediction (CP) provides sets of candidate classes with a guaranteed probability of containing the true class. However, it typically relies on a calibration set with clean labels. We address privacy-sensitive scenarios where the aggregator is untrusted and can only access a perturbed version of the true labels. We propose two complementary approach

Xinbang Dai, Huikang Hu, Yuncheng Hua, Jiaqi Li

Retrieval-augmented generation (RAG) is a promising paradigm, yet its trustworthiness remains a critical concern. A major vulnerability arises prior to generation: models often fail to balance parametric (internal) and retrieved (external) knowledge, particularly when the two sources conflict or are unreliable. To analyze these scenarios comprehensively, we

Chengpu Lv, Jianzhou Zhao, Yueshan Xu, Yu Song

Two-dimensional (2D) massless Dirac fermions (MDF), which represent a type of quasi-particles with linear energy-momentum dispersions only in 2D momentum space, provide a fertile ground for realizing novel quantum phenomena. However, 2D MDF were seldom observed in the superconducting bulk states of 3D materials. Furthermore, as a cornerstone for accurately t

Chen Shani, Liron Soffer, Dan Jurafsky, Yann LeCun

Humans organize knowledge into compact conceptual categories that balance compression with semantic richness. Large Language Models (LLMs) exhibit impressive linguistic abilities, but whether they navigate this same compression-meaning trade-off remains unclear. We apply an Information Bottleneck framework to compare human conceptual structure with embedding

Akash Dhruv, Yangxinyu Xie, Jordan Branham, Tanwi Mallick

This paper presents a comparative study of large language models (LLMs) in interpreting grid-structured geospatial data. We evaluate the performance of a base model through structured prompting and contrast it with a fine-tuned variant trained on a dataset of user-assistant interactions. Our results highlight the strengths and limitations of zero-shot prompt

Richard B. Lehoucq, Scott A. McKinley, Petr Plecháč

In the context of PDE-constrained optimization theory, source identification problems traditionally entail particles emerging from an unknown source distribution inside a domain, moving according to a prescribed stochastic process, e.g.~Brownian motion, and then exiting through the boundary of a compact domain. Given information about the flux of particles t

Mustafa Bozdag, Arya Honarpisheh, Mario Sznaier

This letter presents a density function based safe control synthesis framework for the pursuit-evasion problem. We extend safety analysis to dynamic unsafe sets by formulating a reach-avoid type pursuit-evasion differential game as a robust safe control problem. Using density functions and semi-algebraic set definitions, we derive sufficient conditions for w

Francesca Rizzo

EPW cubes are polarized hyper-K\"ahler varieties of K$3^{[3]}$-type that carry an anti-symplectic involution. We study the geometry of the fixed locus $\sW_A$ of this involution and prove that it is a \emph{rigid} atomic Lagrangian submanifold. Our proof is based on a detailed description of certain singular degenerations of EPW cubes and the degeneration me

Hiromasa Tajima, Riku Yoshimoto, Ryo Nemoto, Yuki Osawa

Entanglement entropy (EE) is widely used to quantify quantum correlations in field theory, with the well-known result in two-dimensional conformal field theory (CFT) predicting a logarithmic divergence with the ultraviolet (UV) cutoff. However, this expression lacks operational meaning: it remains unclear how much of the entanglement is physically extractabl

He Hu, Yucheng Zhou, Juzheng Si, Qianning Wang

Large language models (LLMs) hold significant potential for mental health support, capable of generating empathetic responses and simulating therapeutic conversations. However, existing LLM-based approaches often lack the clinical grounding necessary for real-world psychological counseling, particularly in explicit diagnostic reasoning aligned with standards