May 2025 arXiv papers — page 54

Junwen Yin, Olle Hellman, Samuel Poncé

Charge carrier mobilities are critical parameters in halide perovskite solar cells, governing their average carrier velocity under an applied electric field and overall efficiency. Recent advances in first-principles calculations of electron-phonon interactions and carrier mobilities have enabled predictive computations for perovskite solar cells. However, t

Xinrong Zhao, Puchun Zhou

In this paper, we investigate the prescribed total geodesic curvature problem for generalized circle packing metrics in hyperbolic background geometry on surfaces with infinite cellular decompositions. To address this problem, we introduce a prescribed curvature flow-a discrete analogue of the Ricci flow on noncompact surfaces-specifically adapted to the set

Jens Göbel, Dario Dennstädt, Lukas Lanza, Karl Worthmann

We propose model predictive funnel control, a novel model predictive control (MPC) scheme building upon recent results in funnel control. The latter is a high-gain feedback methodology that achieves evolution of the measured output within predefined error margins. The proposed method dynamically optimizes a parameter-dependent error boundary in a receding-ho

Ruiyi Fang, Bingheng Li, Jingyu Zhao, Ruizhi Pu

Graph Domain Adaptation (GDA) transfers knowledge from labeled source graphs to unlabeled target graphs, addressing the challenge of label scarcity. In this paper, we highlight the significance of graph homophily, a pivotal factor for graph domain alignment, which, however, has long been overlooked in existing approaches. Specifically, our analysis first rev

Nitay Calderon, Liat Ein-Dor, Roi Reichart

Preference mechanisms, such as human preference, LLM-as-a-Judge (LaaJ), and reward models, are central to aligning and evaluating large language models (LLMs). Yet, the underlying concepts that drive these preferences remain poorly understood. In this work, we propose a fully automated method for generating local and global concept-based explanations of pref

Makesh Narsimhan Sreedhar, Traian Rebedea, Christopher Parisien

Reasoning-based language models have demonstrated strong performance across various domains, with the most notable gains seen in mathematical and coding tasks. Recent research has shown that reasoning also offers significant benefits for LLM safety and guardrail applications. In this work, we conduct a comprehensive analysis of training reasoning-based guard

Li Xu, Jiahui Zhang

This paper rigorously analyzes how the {\it large box limit} fundamentally alters the global existence theory and dynamics behavior of the incompressible magnetohydrodynamics (MHD) system with small viscosity/resistivity $(0<\mu\ll 1)$ on periodic domains $Q_L=[-L,L]^3$, in presence of a strong background magnetic field. While the existence of global solutio

Eleonora Cappuccio, Andrea Esposito, Francesco Greco, Giuseppe Desolda

Artificial Intelligence (AI) is one of the major technological advancements of this century, bearing incredible potential for users through AI-powered applications and tools in numerous domains. Being often black-box (i.e., its decision-making process is unintelligible), developers typically resort to eXplainable Artificial Intelligence (XAI) techniques to i

Peng-Xiang Hao, Kotaro Shinmyo, Yu-ki Suzuki, Shunta Takahashi

We explore the reconstruction of bulk local states in three-dimensional flat spacetime (Flat$_3$) using states from two-dimensional Carrollian conformal field theories (CCFT$_2$), proposed as dual field theories in one lower dimension. For massive scalar-type bulk excitations, reconstruction is achieved through states in the induced representation. This meth

Luiz Renato Fontes, Andrea Hernández

We take scaling limits of the Bouchaud and Dean trap model on Parisi's tree in time scales where the dynamics is either ergodic (close to equilibrium) or aging (far from equilibrium). These results follow from a continuity theorem formulated for a certain kind of process on trees, which we call a cascading jump evolution, defined in terms of a collection of

Ryo Ohara, Chi-Lan Yang, Takuji Narumi, Hideaki Kuzuoka

Co-viewing videos with family and friends remotely has become prevalent with the support of communication channels such as text messaging or real-time voice chat. However, current co-viewing platforms often lack visible embodied cues, such as body movements and facial expressions. This absence can reduce emotional engagement and the sense of co-presence when

Yige Yuan, Teng Xiao, Li Yunfan, Bingbing Xu

Aligning large language models with human feedback at inference time has received increasing attention due to its flexibility. Existing methods rely on generating multiple responses from the base policy for search using a reward model, which can be considered as searching in a discrete response space. However, these methods struggle to explore informative ca

Yulin Ye

In this paper, we consider the helicity conservation of weak solutions for the compressible Euler equations in a bounded domain with general pressure law and vacuum. We deduce a sufficient condition for a weak solution conserving the helicity based on the interior Besov-VMO type regularity, the continuous conditions for velocity and vorticity near the bounda

Zaihan Yang, Ryan Leonard, Hien Tran, Rory Driscoll

Suicidal thoughts and behaviors are increasingly recognized as a critical societal concern, highlighting the urgent need for effective tools to enable early detection of suicidal risk. In this work, we develop robust machine learning models that leverage Reddit posts to automatically classify them into four distinct levels of suicide risk severity. We frame

Saurabh Pargal, Abhijit A. Sane

This paper investigates the application of machine learning regression algorithms Kernel Ridge Regression (KRR), Huber Regressor (HR), and Gaussian Process Regression (GPR) for predicting sound power levels of gensets, offering significant value for marketing and sales teams during the early bidding process. When engine sizes and genset enclosure dimensions

Shono Shibuya, Sotaro Sugishita

We consider wave packets of a massless scalar field that have well-localized Rindler energy, and examine how their energy appears to a Minkowski observer to study how the classical gravitational red-shift formula is modified quantum mechanically. We derive, by using the saddle point approximation, an analytic expression for the Minkowski momentum distributio

Davoud Hosseinnezhad, Mel T. Devine, Seán McGarraghy

This paper employs a game-theoretic approach to analyze investment decisions in Ireland's electricity market. It compares optimal electricity investment strategies among energy generators under a perfect competition framework with an imperfect Nash-Cournot competition. The model incorporates market price based on competition among generators while accounting

Jiaye Lin, Mengdi Li, Xufeng Zhao, Wenhao Lu

Reward models trained through Reinforcement Learning from AI Feedback (RLAIF) methods frequently suffer from limited generalizability, which hinders the alignment performance of policy models. This challenge stems from various issues, including distribution shift, preference label noise, and mismatch of overly challenging samples with model capacity. In this

Jinyan Wang, Liu Yang, Yuecen Wei, Jiaxuan Si

Graph Neural Network-based methods face privacy leakage risks due to the introduction of topological structures about the targets, which allows attackers to bypass the target's prior knowledge of the sensitive attributes and realize membership inference attacks (MIA) by observing and analyzing the topology distribution. As privacy concerns grow, the assumpti

D. Melo, L. Landau, R. de Lamare

The next generation of wireless communications systems will employ new frequency bands such as those in the upper midband, millimeter-wave and sub-terahertz frequency bands. The high energy consumption of analog-to-digital converters resulting from their high resolution constituted a major limitation for future wireless communications systems, which will req

Yige Yuan, Teng Xiao, Shuchang Tao, Xue Wang

Large language models (LLMs) have demonstrated impressive performance on reasoning-intensive tasks, but enhancing their reasoning abilities typically relies on either reinforcement learning (RL) with verifiable signals or supervised fine-tuning (SFT) with high-quality long chain-of-thought (CoT) demonstrations, both of which are expensive. In this paper, we

Lukas Meier, Asif I. Bhatti, Leo Kestens, Stefaan Cottenier

Hydrogen uptake into body-centered cubic (bcc) iron as a root cause for subsequent hydrogen embrittlement, is initiated at the surface. In this paper, we quantify how readily H diffuses from the surface into the bulk. We consider a set of low-index, vicinal and general Fe surfaces and treat H-permeation as a two-step process. First, density-functional calcul

Maxime Roumesy, François Ménard, Ryo Tazaki, Gaspard Duchêne

The early stages of planet formation, involving dust grain growth and planetesimals formation, remain shrouded in mystery. The analysis of the Scattering Phase Function (SPF) measured in disks surrounding young stars holds great potential for revealing crucial information about dust grain properties. Given the increasing number of high-quality datasets avail

Leroy Chew, Tomáš Peitl

We show that extension variables in (D)QBF can be generalised by conditioning on universal assignments. The benefit of this is that the dependency sets of such conditioned extension variables can be made smaller to allow easier refutations. This simple modification instantly solves many challenges in p-simulating the QBF expansion rule, which cannot be p-sim

Qingyu Liang, Jaime Banks

Shared understanding plays a key role in the effective communication in and performance of human-human interactions. With the increasingly common integration of AI into human contexts, the future of personal and workplace interactions will likely see human-AI interaction (HAII) in which the perception of shared understanding is important. Existing literature

Isabelle Augenstein, Michiel Bakker, Tanmoy Chakraborty, David Corney

Social media platforms have traditionally relied on internal moderation teams and partnerships with independent fact-checking organizations to identify and flag misleading content. Recently, however, platforms including X (formerly Twitter) and Meta have shifted towards community-driven content moderation by launching their own versions of crowd-sourced fact

Hilde I Hummel, Sandjai Bhulai, Burooj Ghani, Rob van der Mei

The sustainability of the ocean ecosystem is threatened by increased levels of sound pollution, making monitoring crucial to understand its variability and impact. Passive acoustic monitoring (PAM) systems collect a large amount of underwater sound recordings, but the large volume of data makes manual analysis impossible, creating the need for automation. Al

Geon-Hyeong Kim, Yu Jin Kim, Byoungjip Kim, Honglak Lee

As Large Language Models (LLMs) are increasingly deployed in real-world applications, balancing helpfulness and safety has become a central challenge. A natural approach is to incorporate safety constraints into Reinforcement Learning from Human Feedback (RLHF), where recent studies have shown promising progress. However, these methods often rely on auxiliar

Matteo Scandi, Álvaro M. Alhambra

Starting from a microscopic description of weak system-bath interactions, we derive from first principles a quantum master equation that does not rely on the well-known rotating wave approximation. This includes generic many-body systems, with Hamiltonians with vanishingly small energy spacings that forbid that approximation. The equation satisfies a general

Dana Arad, Aaron Mueller, Yonatan Belinkov

Sparse Autoencoders (SAEs) have been proposed as an unsupervised approach to learn a decomposition of a model's latent space. This enables useful applications such as steering - influencing the output of a model towards a desired concept - without requiring labeled data. Current methods identify SAE features to steer by analyzing the input tokens that activa

Michael Günther, Adrian Sandu

This survey provides an overview of state-of-the art multirate schemes, which exploit the different time scales in the dynamics of a differential equation model by adapting the computational costs to different activity levels of the system. We start the discussion with the straightforward approach based on interpolating and extrapolating the slow--fast coupl

Radhakrishna V, Anurag Tyagi, Koushal Vadodariya, Vivek K Agrawal

X-ray Polarimeter Satellite (XPoSat) with POLarimeter Instrument in X-rays (POLIX), is India's first spacecraft dedicated to study medium energy X-ray polarisation from celestial objects. X-Ray Spectroscopy and Timing (XSPECT) instrument on XPoSat is configured to study long term spectral behaviour of select sources in Soft X-ray regime. The instrument uses

Abraão Mendes

This paper investigates the geometric consequences of equality in area-charge inequalities for spherical minimal surfaces and, more generally, for marginally outer trapped surfaces (MOTS), within the framework of the Einstein-Maxwell equations. We show that, under appropriate energy and curvature conditions, saturation of the inequality $\mathcal{A} \geq 4\p

Chang Sun, Hui Yuan, Shiqi Jiang, Da Ai

Since the data volume of LiDAR point clouds is very huge, efficient compression is necessary to reduce their storage and transmission costs. However, existing learning-based compression methods do not exploit the inherent angular resolution of LiDAR and ignore the significant differences in the correlation of geometry information at different bitrates. The p

Huiyao Chen, Yi Yang, Yinghui Li, Meishan Zhang

Existing long-document question answering systems typically process texts as flat sequences or use heuristic chunking, which overlook the discourse structures that naturally guide human comprehension. We present a discourse-aware hierarchical framework that leverages rhetorical structure theory (RST) for long document question answering. Our approach convert

Yihong Lin, Xianjia Wu, Xilai Wang, Jianqiao Hu

Monocular 3D hand mesh recovery is challenging due to high degrees of freedom of hands, 2D-to-3D ambiguity and self-occlusion. Most existing methods are either inefficient or less straightforward for predicting the position of 3D mesh vertices. Thus, we propose a new pipeline called Monocular 3D Hand Mesh Recovery (M3DHMR) to directly estimate the positions

Alex Bishop, Murray Elder, Alex Evetts, Paul Gallot

We prove that the word problem for the infinite cyclic group is not EDT0L, and obtain as a corollary that a finitely generated group with EDT0L word problem must be torsion. In addition, we show that the property of having an EDT0L word problem is invariant under change of generating set and passing to finitely generated subgroups. This represents significan

Hongsong Wang, Yin Zhu, Qiuxia Lai, Yang Zhang

Computational dance generation is crucial in many areas, such as art, human-computer interaction, virtual reality, and digital entertainment, particularly for generating coherent and expressive long dance sequences. Diffusion-based music-to-dance generation has made significant progress, yet existing methods still struggle to produce physically plausible mot

James Merrick, Lewis Hutton, Joseph C. Cooper, Claire Vallance

The effect of spin-orbit coupling on the "Newton's cradle"-type photodynamics in the cyclic disulfide 1,2-dithiane (C4H8S2) is investigated theoretically. We consider excitation by a 290 nm laser pulse and simulate the subsequent ultrafast nonadiabatic dynamics by propagating surface-hopping trajectories using SA(4|4)-CASSCF(6,4)-level electronic structure c

Francesco De Pas, Serena Dipierro, Mirco Piccinini, Enrico Valdinoci

We investigate existence, uniqueness and asymptotic behavior of minimizers of a family of non-local energy functionals of the type $$ \frac{1}{4}\iint_{\mathbb{R}^{2n}\setminus (\mathbb{R}^n \setminus \Omega)^2}|u(x)-u(y)|^2 K(x-y) \,dx dy + \int_\Omega W(u(x)) \,dx. $$ Here, $W$ is a possibly degenerate double well potential with a polynomial control on its

Zheqi Lv, Junhao Chen, Qi Tian, Keting Yin

Diffusion models have become the mainstream architecture for text-to-image generation, achieving remarkable progress in visual quality and prompt controllability. However, current inference pipelines generally lack interpretable semantic supervision and correction mechanisms throughout the denoising process. Most existing approaches rely solely on post-hoc s

Hazem Alsamkary, Mohamed Elshaffei, Mohamed Elkerdawy, Ahmed Elnaggar

Protein language models (PLMs) have emerged as powerful tools to detect complex patterns of protein sequences. However, the capability of PLMs to fully capture information on protein sequences might be limited by focusing on single pre-training tasks. Although adding data modalities or supervised objectives can improve the performance of PLMs, pre-training o

Gianmarco Genalti, Sujay Bhatt, Nicola Gatti, Alberto Maria Metelli

Regret minimization in stochastic non-stationary bandits gained popularity over the last decade, as it can model a broad class of real-world problems, from advertising to recommendation systems. Existing literature relies on various assumptions about the reward-generating process, such as Bernoulli or subgaussian rewards. However, in settings such as finance

Alkis Koudounas, Moreno La Quatra, Gabriele Ciravegna, Marco Fantini

Voice disorders significantly impact patient quality of life, yet non-invasive automated diagnosis remains under-explored due to both the scarcity of pathological voice data, and the variability in recording sources. This work introduces MVP (Multi-source Voice Pathology detection), a novel approach that leverages transformers operating directly on raw voice

Hongsong Wang, Ao Sun, Jie Gui, Liang Wang

Gesture recognition is an important research area in the field of computer vision. Most gesture recognition efforts focus on close-set scenarios, thereby limiting the capacity to effectively handle unseen or novel gestures. We aim to address class-incremental gesture recognition, which entails the ability to accommodate new and previously unseen gestures ove

Ali Forootani, Mohammad Khosravi

Time series forecasting plays a critical role in domains such as energy, finance, and healthcare, where accurate predictions inform decision-making under uncertainty. Although Transformer-based models have demonstrated success in sequential modeling, their adoption for time series remains limited by challenges such as noise sensitivity, long-range dependenci

Debargha Ganguly, Vikash Singh, Sreehari Sankar, Biyao Zhang

Large language models (LLMs) show remarkable promise for democratizing automated reasoning by generating formal specifications. However, a fundamental tension exists: LLMs are probabilistic, while formal verification demands deterministic guarantees. This paper addresses this epistemological gap by comprehensively investigating failure modes and uncertainty

Le Zhang, Bo Wang, Xipeng Qiu, Siva Reddy

We present REARANK, a large language model (LLM)-based listwise reasoning reranking agent. REARANK explicitly reasons before reranking, significantly improving both performance and interpretability. Leveraging reinforcement learning and data augmentation, REARANK achieves substantial improvements over baseline models across popular information retrieval benc

Puxian Wei, Ruifeng Zheng, Qiaoli Yang

Both axion and dark photon dark matter are among the most promising candidates of dark matter. What we know with some confidence is that they exhibit a small velocity distribution $\delta v\lesssim v\sim 10^{-3}$c. In addition, their mass is small, resulting in a long de Broglie wavelength and a high particle number density. Their phase space distribution co

Marcello Cellina, Matteo Corno, Sergio Matteo Savaresi

High Speed multi-vehicle Autonomous Racing will increase the safety and performance of road-going Autonomous Vehicles. Precise vehicle detection and dynamics estimation from a moving platform is a key requirement for planning and executing complex autonomous overtaking maneuvers. To address this requirement, we have developed a Latency-Aware EKF-based Multi

Reinis Irmejs, Mari Carmen Bañuls, J. Ignacio Cirac

We investigate the performance of an adiabatic evolution protocol when initialized from a Gibbs state at finite temperature. Specifically, we identify the diagonality of the final state in the energy eigenbasis, as well as the difference in energy and in energy variance with respect to the ideal adiabatic limit as key benchmarks for success and introduce met

Jianxin Huang, Jiahang Li, Sergey Vityazev, Alexander Dvorkovich

RGB-D scene parsing methods effectively capture both semantic and geometric features of the environment, demonstrating great potential under challenging conditions such as extreme weather and low lighting. However, existing RGB-D scene parsing methods predominantly rely on supervised training strategies, which require a large amount of manually annotated pix

Yasutaka Koga, Ryota Maeda, Daiki Saito, Keiya Uemichi

We construct static, spherically symmetric, charged traversable wormhole solutions to the Einstein--Maxwell equations, supported by bidirectional (ingoing and outgoing) null dust with negative energy, and discuss a scenario for their dynamical formation from a black hole. Our solution contains a traversable throat, where the areal radius takes a minimum, alt

Marino Coppolaro, Massimo Moccia, Giuseppe Castaldi, Vincenzo Galdi

We present a theoretical framework for analyzing aperiodically ordered photonic time quasicrystals (PTQCs), which are the temporal analogs of spatial photonic quasicrystals. Using a general two-symbol substitutional sequence to model temporal modulations, we extend the trace and anti-trace map formalism used for spatial photonic quasicrystals to the temporal

Chang Liu, Haomin Zhang, Shiyu Xia, Zihao Chen

Generating high-quality piano audio from video requires precise synchronization between visual cues and musical output, ensuring accurate semantic and temporal alignment.However, existing evaluation datasets do not fully capture the intricate synchronization required for piano music generation. A comprehensive benchmark is essential for two primary reasons:

Abhiram Sripat

Let X be a connected Hausdorff surface, let Sigma be a finite subset of X, and let pi from Y to X minus Sigma be a finite sheeted covering. Peripheral monodromy at each marked point determines a finite set of peripheral component germs together with their local degree labels. We introduce incidence completions by adjoining an exceptional fibre at each marked

Andrei V. Korol, Andrey V. Solov'yov

In this paper, theoretical and numerical analyses are conducted of the profiles of the planar (-110) crystallographic direction in the diamond layer doped with boron atoms. The planar profiles for periodic doping following several ideal dependencies of the boron concentration on the distance in the crystalline medium. Numerical simulations of the channeling

Hazem Alsamkary, Mohamed Elshaffei, Mohamed Soudy, Sara Ossman

Protein-protein interactions (PPIs) are fundamental to numerous cellular processes, and their characterization is vital for understanding disease mechanisms and guiding drug discovery. While protein language models (PLMs) have demonstrated remarkable success in predicting protein structure and function, their application to sequence-based PPI binding affinit

A. S. Bonomo, L. Naponiello, E. Pezzetta, A. Sozzetti

Determining whether there is any correlation between the presence of short-period small planets (SPs) with $P\lesssim100~d$ ($a \lesssim0.4~AU$) and $1<M_{p}<20~M_\oplus$ and that of outer cold Jupiters (CJs) with $a=1-10~AU$ and $M_{p}=0.5-20~M_{Jup}$ around solar-type stars may provide crucial constraints on models of formation and/or migration of SPs. How

Liulu He, Shenli Zheng, Karwei Sun, Yijiang Liu

Rotations have become essential to state-of-the-art quantization pipelines for large language models (LLMs) by effectively smoothing outliers in weights and activations. However, further optimizing the rotation parameters offers only limited performance gains and introduces significant training overhead: due to rotation parameter sharing, full-model must be

Juwei Yue, Haikuo Li, Jiawei Sheng, Yihan Guo

Dynamics modeling has been introduced as a novel paradigm in message passing (MP) of graph neural networks (GNNs). Existing methods consider MP between nodes as a heat diffusion process, and leverage heat equation to model the temporal evolution of nodes in the embedding space. However, heat equation can hardly depict the wave nature of graph signals in grap

Agam Shayit, Can Liao, Shiv Upadhyay, Hang Hu

The combinatorial scaling of configuration interaction (CI) has long restricted its applicability to only the simplest molecular systems. Here, we report the first numerically exact CI calculation exceeding one quadrillion ($10^{15}$) determinants, enabled by categorical compression within the small-tensor-product distributed active space (STP-DAS) framework

Christoph Schuhmann, Robert Kaczmarczyk, Gollam Rabby, Felix Friedrich

Effective human-AI interaction relies on AI's ability to accurately perceive and interpret human emotions. Current benchmarks for vision and vision-language models are severely limited, offering a narrow emotional spectrum that overlooks nuanced states (e.g., bitterness, intoxication) and fails to distinguish subtle differences between related feelings (e.g.

Fotios Lygerakis, Ozan Özdenizci, Elmar Rückert

Tactile sensing provides local essential information that is complementary to visual perception, such as texture, compliance, and force. Despite recent advances in visuotactile representation learning, challenges remain in fusing these modalities and generalizing across tasks and environments without heavy reliance on pre-trained vision-language models. More

Uktamjon Uktamov, Sanjar Shaymatov, Bobomurat Ahmedov, Chengxun Yuan

A novel analytical Schwarzschild-like black hole (BH) solution is derived. It exhibits a static BH with a dark matter (DM) halo characterized by a Dehnen-type density profile. This solution could represent an alternative perspective on the interaction of black hole-dark matter systems, providing new insights into the fundamental properties of DM halos. We st

Subba Reddy Oota, Akshett Jindal, Ishani Mondal, Khushbu Pahwa

Transformer-based language models, though not explicitly trained to mimic brain recordings, have demonstrated surprising alignment with brain activity. Progress in these models-through increased size, instruction-tuning, and multimodality-has led to better representational alignment with neural data. Recently, a new class of instruction-tuned multimodal LLMs

Florian Raßhofer, Simon Bauer, Alexander Ziepke, Ivan Maryshev

The dynamics of phase-separated interfaces shape the behavior of both passive and active condensates. While surface tension in equilibrium systems minimizes interface length, non-equilibrium fluxes can destabilize flat or constantly curved interfaces, giving rise to complex interface morphologies. Starting from a minimal model that couples a conserved, phase

Subba Reddy Oota, Khushbu Pahwa, Mounika Marreddy, Maneesh Singh

Despite participants engaging in unimodal stimuli, such as watching images or silent videos, recent work has demonstrated that multi-modal Transformer models can predict visual brain activity impressively well, even with incongruent modality representations. This raises the question of how accurately these multi-modal models can predict brain activity when p

Xinping Chen, Chen Liu

We propose Gradient Inversion Transcript (GIT), a novel generative approach for reconstructing training data from leaked gradients. GIT employs a generative attack model, whose architecture is tailored to align with the structure of the leaked model based on theoretical analysis. Once trained offline, GIT can be deployed efficiently and only relies on the le

Łukasz Merta, Filip Zieliński, Marcin Zieliński

We give a complete classification of free arrangement of three smooth conics on complex projective plane admitting only ${\rm ADE}$ singularities and $J_{2,0}$ singularities.

Xueyi Liu, Zuodong Zhong, Yuxin Guo, Yun-Fu Liu

Due to the powerful vision-language reasoning and generalization abilities, multimodal large language models (MLLMs) have garnered significant attention in the field of end-to-end (E2E) autonomous driving. However, their application to closed-loop systems remains underexplored, and current MLLM-based methods have not shown clear superiority to mainstream E2E

Yihan Chen, Benfeng Xu, Xiaorui Wang, Yongdong Zhang

Autonomous agents, which perceive environments and take actions to achieve goals, have become increasingly feasible with the advancements in large language models (LLMs). However, current powerful agents often depend on sophisticated prompt engineering combined with closed-source LLMs like GPT-4. Although training open-source LLMs using expert trajectories f

Xin Bing, Xin He, Chao Wang

Kernel methods, particularly kernel ridge regression (KRR), are time-proven, powerful nonparametric regression techniques known for their rich capacity, analytical simplicity, and computational tractability. The analysis of their predictive performance has received continuous attention for more than two decades. However, in many modern regression problems wh

Hyunsik Chae, Seungwoo Yoon, Jaden Park, Chloe Yewon Chun

Recent Vision-Language Models (VLMs) have demonstrated impressive multimodal comprehension and reasoning capabilities, yet they often struggle with trivially simple visual tasks. In this work, we focus on the domain of basic 2D Euclidean geometry and systematically categorize the fundamental, indivisible visual perception skills, which we refer to as atomic

Natallia Kokash, Lei Wang, Thomas H. Gillespie, Adam Belloum

The rise of electronic health records (EHRs) has unlocked new opportunities for medical research, but privacy regulations and data heterogeneity remain key barriers to large-scale machine learning. Federated learning (FL) enables collaborative modeling without sharing raw data, yet faces challenges in harmonizing diverse clinical datasets. This paper present

A. Manca, A. Marino, A. Borghese, F. Coti Zelati

MAXI J1834-021 is a new X-ray transient that was discovered in February 2023. We analysed the spectral and timing properties of MAXI J1834-021 using NICER, NuStar and Swift data collected between March and October 2023. The light curve showed a main peak followed by a second activity phase. The majority of the spectra extracted from the individual NICER obse

Jonathan Caalim, Yu-ichi Tanaka

A sequence $(e_i)_{i \le m}$ of nonnegative integers $e_i$, where $m \in \mathbb{N}$ or $m =\infty$, is called a binomid index if $\sum_{i=n-k+1}^{n} e_i\geq \sum_{i=1}^ke_i$ for all $k, n \in \mathbb{N}$ such that $ 1\le k \le n < m$. Infinite binomid indices give rise to binomid sequences (also known as Raney sequences) and generalized binomial coefficient

Brady D. Lund, Ting Wang

This paper reviews literature pertaining to the development of data science as a discipline, current issues with data bias and ethics, and the role that the discipline of information science may play in addressing these concerns. Information science research and researchers have much to offer for data science, owing to their background as transdisciplinary s

Baptiste Abélès, Eugenio Clerico, Hamish Flynn, Gergely Neu

We study the linear stochastic bandit problem, relaxing the standard i.i.d. assumption on the observation noise. As an alternative to this restrictive assumption, we allow the noise terms across rounds to be sub-Gaussian but interdependent, with dependencies that decay over time. To address this setting, we develop new confidence sequences using a recently i

Chengrui Huang, Shen Gao, Zhengliang Shi, Dongsheng Wang

Existing tool-learning methods usually rely on supervised fine-tuning, they often overlook fine-grained optimization of internal tool call details, leading to limitations in preference alignment and error discrimination. To overcome these challenges, we propose Token-level Tool-use Preference Alignment Training Framework (TTPA), a training paradigm for const

Ramon Ferrer-i-Cancho

Here we present a new class of optimality for coding systems. Members of that class are displaced linearly from optimal coding and thus exhibit Zipf's law, namely a power-law distribution of frequency ranks. Within that class, Zipf's law, the size-rank law and the size-probability law form a group-like structure. We identify human languages that are members

Hoyun Song, Huije Lee, Jisu Shin, Sukmin Cho

The detection of mental health problems from social media and the interpretation of these results have been extensively explored. Research has shown that incorporating clinical symptom information into a model enhances domain expertise, improving its detection and interpretation performance. While large language models (LLMs) are shown to be effective for ge

Minda Hu, Tianqing Fang, Jianshu Zhang, Junyu Ma

Web agents powered by Large Language Models (LLMs) show promise for next-generation AI, but their limited reasoning in uncertain, dynamic web environments hinders robust deployment. In this paper, we identify key reasoning skills essential for effective web agents, i.e., reflection & lookahead, branching, and rollback, and curate trajectory data that exempli

Mary John, Sherzod Turaev, Saja Al-Dabet, Rawad Abdulghafor

This study presents a comprehensive framework for evaluating the quality of public spaces across various urban typologies. Through a systematic review of 159 research studies, we identify universal quality factors that transcend spatial types as well as specialized factors unique to specific public environments. Our findings establish accessibility (73.6%),

Xingyu Xiao, Jiejuan Tong, Jun Sun, Zhe Sui

Digitalization in nuclear power plant (NPP) control rooms is reshaping how operators interact with procedures and interface elements. However, existing computer-based procedures (CBPs) often lack semantic integration with human-system interfaces (HSIs), limiting their capacity to support intelligent automation and increasing the risk of human error, particul

Yu Yao, Junhao Zhang, Pu Miao, Long Zhang

Unlike the conventional fully-passive and fully-active reconfigurable intelligent surfaces (RISs), a hybrid RIS consisting of active and passive reflection units has recently been concerned, which can exploit their integrated advantages to alleviate the RIS-induced path loss. In this paper, we investigate a novel security strategy where the multiple hybrid R

Maciej Swiechowski, Dominik Slezak

Human-like agents are an increasingly important topic in games and beyond. Believable non-player characters enhance the gaming experience by improving immersion and providing entertainment. They also offer players the opportunity to engage with AI entities that can function as opponents, teachers, or cooperating partners. Additionally, in games where bots ar

Gianmarco Genalti, Francesco Emanuele Stradi, Matteo Castiglioni, Alberto Marchesi

This paper initiates the study of data-dependent regret bounds in constrained MAB settings. These bounds depend on the sequence of losses that characterize the problem instance. Thus, they can be much smaller than classical $\widetilde{\mathcal{O}}(\sqrt{T})$ regret bounds, while being equivalent to them in the worst case. Despite this, data-dependent regret

Victoria Barrett, Alessio Guglielmi, Benjamin Ralph, Lutz Straßburger

Subatomic logic is a recent innovation in structural proof theory where atoms are no longer the smallest entity in a logical formula, but are instead treated as binary connectives. As a consequence, we can give a subatomic proof system for propositional classical logic such that all derivations are strictly linear: no inference step deletes or adds informati

Didier Lairez

Energy has no definition, except that given by a conservation principle which essentially amounts to defining it as the elements of an open list of unknown cardinality. Entropy, identified by Shannon as information we lack, has too many definitions. This results in an unstable and hesitant interpretation of their link. Thermodynamics, the science of changes

Benoit Dagallier, Claudio Landim

We consider a one-dimensional microscopic reaction-diffusion process obtained as a superposition of a Glauber and a Kawasaki dynamics. The reaction term is tuned so that a dynamical phase transition occurs in the model as a suitable parameter is varied. We study dynamical fluctuations of the density field at the critical point. We characterise the slowdown o

Lucas Ueda, João Lima, Leonardo Marques, Paula Costa

Emotion plays a fundamental role in human interaction, and therefore systems capable of identifying emotions in speech are crucial in the context of human-computer interaction. Speech emotion recognition (SER) is a challenging problem, particularly in natural speech and when the available data is imbalanced across emotions. This paper presents our proposed s

Raphaël Bagat, Irina Illina, Emmanuel Vincent

We aim to improve the robustness of Automatic Speech Recognition (ASR) systems against non-native speech, particularly in low-resourced multi-accent settings. We introduce Mixture of Accent-Specific LoRAs (MAS-LoRA), a fine-tuning method that leverages a mixture of Low-Rank Adaptation (LoRA) experts, each specialized in a specific accent. This method can be

Jack Storror Carter

Penalised likelihoods are often used for sparse estimation of a Gaussian precision matrix. In high dimensional settings where the matrix dimension is larger than the sample size, the sample covariance matrix $S$ is not of full rank and the maximum likelihood estimate of the precision matrix does not exist. An additional advantage of some penalised likelihood

Rui-Jing Wang

In this paper, it is proved that there is an arithmetic progression of positive integers such that each of which is expressible neither as $p+F_m$ nor as $q+L_n$, where $ p,q $ are primes, $ F_m $ denotes the $ m $-th Fibonacci number and $ L_n $ denotes the $ n $-th Lucas number.

Rongqi Pan, Feifei Niu, Lionel C. Briand, Hanyang Hu

As software systems evolve, test suites tend to grow in size and often contain redundant test cases. Such redundancy increases testing effort, time, and cost. Test suite minimization (TSM) aims to eliminate such redundancy while preserving key properties such as requirement coverage and fault detection capability. In this paper, we propose RTM (Requirement c

Qiong Zhang, Yan Shuo Tan, Qinglong Tian, Pengfei Li

Hollmann et al. (Nature 637 (2025) 319-326) recently introduced TabPFN, a transformer-based deep learning model for regression and classification on tabular data, which they claim "outperforms all previous methods on datasets with up to 10,000 samples by a wide margin, using substantially less training time." Furthermore, they have called TabPFN a "foundatio

G. Giunta, M. Gorgone, F. Oliveri

A network of agents interacting both with competitive and/or cooperative mechanisms is modeled by using fermionic ladder operators. The time evolution of the network is assumed to be governed by a Hermitian time-independent Hamiltonian operator, and the mean values of the number operators are interpreted as a measure of the wealth status of the agents. Besid

Ondŕej Straka, Uwe D. Hanebeck

Gaussian mixture filters for nonlinear systems usually rely on severe approximations when calculating mixtures in the prediction and filtering step. Thus, offline approximations of noise densities by Gaussian mixture densities to reduce the approximation error have been proposed. This results in exponential growth in the number of components, requiring ongoi

Shihao Li, Huaibo Huang, Junxian Duan, Aihua Zheng

Multi-modal object Re-IDentification (ReID) aims to obtain complete identity features across heterogeneous modalities. However, most existing methods rely on implicit feature fusion modules, making it difficult to model fine-grained recognition patterns under various challenges in real world. Benefiting from the powerful Multi-modal Large Language Models (ML