April 2026 arXiv papers — page 142

Ilhuan Choi, Jiwon Yoo, Yoona Lee, Yewon Jeong

Write-and-verify (WV) is essential for programming multi-level RRAM weights, yet under scaled-voltage and low-SNR conditions the verify read increasingly limits mapping accuracy, convergence speed and energy. We propose a Hadamard-domain WV framework that improves verify reliability without adding analog hardware. % without introducing additional analog bloc

Chieh Tsai, Hossein Rastgoftar, Salim Hariri

Autonomous vehicles are increasingly deployed in safety-critical applications, where sensing failures or cyberphysical attacks can lead to unsafe operations resulting in human loss and/or severe physical damages. Reliable real-time perception is therefore critically important for their safe operations and acceptability. For example, vision-based distance est

Kimon Boehmer

In the problem of Submodular Max-Min Allocation, we are given a set of items, a set of players, and monotone submodular valuation functions that represent the satisfaction of a player with a certain subset of items. The goal is to find an allocation of the items to the players that maximizes the lowest satisfaction among all players. We study this problem in

Kaito Ito, Anqi Dong

Optimal transport with quadratic cost provides a geometric framework for steering an ensemble, modeled by a probability law, with minimal effort. Yet ambient-space formulations become unwieldy in high dimensions, and sensing or actuation in practice often reveals only linear views of the state -- camera silhouettes, LiDAR beams, tomographic slices. We develo

Urs Wenger

In this review I summarize how machine learning can be used in lattice gauge theory simulations and what ap\-proaches are currently available to improve the sampling of gauge field configurations, with a focus on applications in four-dimensional SU(3) gauge theories. These include approaches based on generative machine-learning models such as (stochastic) no

Giuseppe Antonio Veltri

We study the existence and localization of eigenvalue-eigenfunction pairs for parameter-dependent Neumann BVPs with a functional term. By reformulating the problems as a Hammerstein integral equation, we apply an existence and localization result and propose a convergent fixed-point iteration scheme. Finally, two pseudocodes and a MATLAB implementation are p

Shiro Mukae, Erin Mentuch Cooper, Karl Gebhardt, Dustin Davis

We present a deep learning framework to enhance the identification of Ly$\alpha$ emitters (LAEs) in the Hobby-Eberly Telescope Dark Energy Experiment (HETDEX), an untargeted spectroscopic survey of LAEs at $1.9 < z < 3.5$ without imaging pre-selection. We primarily address the low signal-to-noise ratio (S/N) regime ($4.8 \leq \mathrm{S/N} \leq 5.5$), where L

Lunbing Chen, Jixin Lu, Yufei Yin, Jinpeng Huang

Dynamic soaring enables sustained flight by extracting energy from wind shear, yet it is commonly understood as a cycle-level maneuver that assumes stable flow conditions. In realistic unsteady environments, however, such assumptions are often violated, raising the question of whether explicit cycle-level planning is necessary. Here, we show that dynamic soa

Mikołaj Sienicki, Krzysztof Sienicki

Roman Schnabel's article argues that the Einstein-Podolsky-Rosen (EPR) paradox can be resolved by identifying a flaw in what the author calls the "EPR implication" and by using radioactive alpha decay as an example showing that predictability does not exclude genuine randomness. The paper is clearly written and addresses an important foundational question. I

Abdourahmane Diatta, El Hadji Abdoulaye Thiam

Let $N \ge 4$, $\Omega$ be a bounded domain in $\mathbb{R}^N$, and let $\Sigma \subset \Omega$ be a smooth closed submanifold of dimension $k$ with $2 \le k \le N-2$. We study the existence of positive solutions $u \in H_0^1(\Omega)$ to the Euler--Lagrange equation \[ -\Delta u + h u = \lambda\, \rho_{\Sigma}^{-s_1}\, u^{2^{*}_{s_1}-1} + \rho_{\Sigma}^{-s_2}

Qiuyu Tian, Haoliang Sun, Yunshan Wang, Yinghuan Shi

Segmentation models based on deep neural networks demonstrate strong generalization for medical image segmentation. However, they often exhibit overconfidence or underconfidence, leading to unreliable confidence scores for segmentation masks, especially in ambiguous regions. This undermines the trustworthiness required for clinical deployment. Motivated by t

Andrey R. Kolovsky

This chapter gives an overview of transport problems where chaotic dynamics of the system plays a crucial role. We begin with single-particle transport problems and then come to conservative and then dissipative systems of identical particles, which follows the historical way of developing the theory of Quantum Chaos over the past 40 years. We also include b

Chieh Tsai, Murad Mehrab Abrar, Salim Hariri

Autonomous vehicles (AVs) promise efficient, clean and cost-effective transportation systems, but their reliance on sensors, wireless communications, and decision-making systems makes them vulnerable to cyberattacks and physical threats. This chapter presents novel design techniques to strengthen the security and resilience of AVs. We first provide a taxonom

Gregory Morse, Tamás Kozsik

Classical computability theory tells us that self-modifying code (SMC) on a deterministic universal Turing machine can be simulated by non-SMC code on the same model. That abstraction, however, omits the external timing inputs, concurrency, and microarchitectural state that dominate practical execution on modern processors. We argue that once timing, orderin

Mirco Lescart, Anna Kiriliouk, Philippe Naveau

Extreme value theory offers a statistical framework for quantifying the risk of rare events, with the generalized Pareto (GP) distribution providing the canonical limit model for univariate threshold exceedances. In many applications, however, extremes are intrinsically multivariate, requiring models that capture both marginal tail behaviours and joint extre

Jiangdong Ai, Huiqiu Lin, Yongtang Shi

We study the first nonzero Steklov eigenvalue $\lambda_2(T,\delta\Omega)$ of the Dirichlet-to-Neumann operator on a finite tree $T$ with leaf boundary $\delta\Omega$, under a constraint on the diameter $D$. He and Hua [Calc. Var. PDE, 2022] showed that $\lambda_2(T) \leq 2/D$ for any tree of diameter $D$, with the even-diameter equality case fully characteri

Jungwon Choi, Eunwoo Kim

Test-Time Prompt Tuning (TPT) adapts vision-language models using augmented views, but its effectiveness is hindered by the challenge of determining which views are beneficial. Standard entropy-based filtering relies on the internal confidence scores of the model, which are often miscalibrated under distribution shift, assigning high confidence to irrelevant

Begum Atesli, Ogul Esen, Michal Pavelka

We introduce a new geometric framework for relativistic particle dynamics based on contact geometry and suitable for treating dissipative processes like particle decay. The dynamics is formulated on a nine--dimensional extended phase space consisting of four position coordinates, four momenta, and an additional variable (functioning as a geometric variant of

Zhijie Cai, Yuhao Zheng, Haolong Chen, Dongzhu Liu

Federated Learning (FL) offers a promising pathway for collaboratively fine-tuning Large Language Models (LLMs) at the edge; however, this paradigm faces a critical bottleneck: the prohibitive communication and memory overheads incurred by exchanging high-dimensional gradients. Furthermore, recent studies reveal that user training data can still be recovered

Xinghua Sun, Wenhai Lin, Ruike Zhou

The performance analysis of wireless CSMA networks is notoriously difficult due to the intricate sensing and interference relationships among links. Even the fundamental problem of throughput characterization remains open when sensing and interference topologies are both arbitrary. In this paper, we develop a new analytical framework for throughput character

Xudong Wang, Chaoning Zhang, Chenghao Li, Shuxu Chen

Large Language Models (LLMs) have demonstrated strong capabilities in complex reasoning tasks, while recent prompting strategies such as Chain-of-Thought (CoT) have further elevated their performance in handling complex logical problems. Despite these advances, high-quality reasoning remains heavily reliant on manual static prompts and is sensitive to decodi

Tao-Zhi Yang, Zhao-Yu Zuo

We report a photometric and asteroseismic analysis of KIC 3868420, a newly identified high-amplitude $\delta$ Scuti-$\gamma$ Doradus hybrid star located in the Hertzsprung gap - a short-lived and rarely observed post-main-sequence phase. Using four years of Kepler long-cadence photometry, we detect 36 significant frequencies, including 11 independent modes s

Sashi Novitasari, Takashi Fukuda, Kurata Gakuto, George Saon

Speech-aware LLMs (SLLMs) have recently achieved state-of-the-art ASR performance; however, they still fail to accurately transcribe bias words that appear rarely or never in the training data. Contextual biasing mechanisms are commonly implemented by introducing a predefined bias word list into the model via a text prompt or additional module. For further i

Yudong Li, Jiawei Cai, Linlin Shen

Standard Large Language Model (LLM) pre-training typically treats corpora as flattened token sequences, often overlooking the real-world context that humans naturally rely on to contextualize information. To bridge this gap, we introduce Knowledge Coordinate Conditioning (KoCo), a simple method that maps every document into a three-dimensional semantic coord

Ayush Agrawal, Aparna Bansal, D. N. Pandey

We study the Stokes--Poisson--Boltzmann equations with Dirichlet and Navier boundary conditions. The system consists of the incompressible Stokes equations coupled with a nonlinear Poisson--Boltzmann equation through electrostatic forcing and convective transport effects. To handle the Navier boundary conditions in a unified framework, we employ Nitsche's me

Sricharan Raghavan-Chitra, Arghadip Koner, Joel Yuen-Zhou

Linear optical spectra of molecular aggregates are often approximated by classical optics methods such as the discrete-dipole approximation (DDA), coherent exciton scattering (CES), and coherent potential approximation (CPA), where the only quantum-mechanical input to the calculation is the linear susceptibility of the monomers. However, the limits of validi

Shuanghong Li, Yongliang Song, Kun Wang, Xianyong Bai

White-light flares (WLFs) are crucial for understanding the energy transport and heating processes in the lower solar atmosphere. Systematic studies are highly necessary. However, most WLFs are very weak and difficult to detect. To address this, we propose a new method of detecting WLFs. Through the observations of SDO/HMI, the light curve of each pixel in t

N. I. Stoilova, J. Van der Jeugt

In this paper we reexamine the definition of parafermions and parabosons by means of Green's triple relations, and extend these relations by including a parity operator $P$ which is also determined by means of triple relations. As a consequence, we are dealing with new algebraic structures. It is shown that the algebra underlying a set of $n$ parafermions to

Jean-Luc Baril, Aubrey Blecher, José Luis ramírez

Stanley polyominoes are a subclass of parallelogram polyominoes in which each row begins strictly to the right of the beginning of the previous row and ends strictly to the right of the end of the previous row. In this paper, we derive generating functions for Stanley polyominoes based on the numbers of columns and rows, area, semiperimeter, and numbers of i

Jiawei Fan, Shigeng Wang, Chao Li, Xiaolong Liu

In this paper, we present Chain-of-Models Pre-Training (CoM-PT), a novel performance-lossless training acceleration method for vision foundation models (VFMs). This approach fundamentally differs from existing acceleration methods in its core motivation: rather than optimizing each model individually, CoM-PT is designed to accelerate the training pipeline at

Lei Lin, Jizhao Zhu, Yong Liu, Donghong Sun

This paper addresses two limitations of large language models (LLMs) in solving complex problems: (1) their reasoning processes exhibit Bayesian-like stochastic generation, where each token is sampled from a context-dependent probability distribution, leading to inherently random decision trajectories rather than deterministic planning; (2) the reasoning and

Zhe Zhang, Yigitcan Özer, Junichi Yamagishi

Speech audio in the wild is often processed by post-production effects, but existing speech datasets rarely provide precise annotations of effects and parameters, limiting systematic study. We introduce VoxEffects, a speech audio effects dataset that pairs produced speech with exact effect-chain supervision at multiple granularities. VoxEffects supports spee

Michael Mandl

The complex Langevin approach is a promising method for the numerical treatment of systems with a sign problem, for which conventional lattice field theory techniques based on importance sampling cannot be applied. However, complex Langevin dynamics may fail to converge in some cases and converge to a wrong limit in others, motivating the development of vari

Yun Peng, Yujun Sun, Jia Ding, Bin Yan

Command-line bioinformatics tools remain essential for genomic analysis, yet their diversity in syntax and parameterization presents a persistent barrier to productive research. We present oxo-call, a Rust-based command-line assistant that translates natural-language task descriptions into accurate tool invocations through two complementary strategies: docum

Andrea Bonvicini, Sonia Coriani, Benoît Champagne

The theory of hyper-Rayleigh scattering optical activity (HRS-OA) spectroscopy has previously been described within the length formulation of the pure electric-dipole and mixed (electric-dipole/magnetic-dipole and electric-dipole/electric-quadrupole) first hyperpolarizabilities required for the description of this process. In this work, we provide an alterna

Jiarui Zhang, Xiangyu Liu, Yong Hu, Chaoyue Niu

Multi-turn dialogue is the predominant form of interaction with large language models (LLMs). While LLM routing is effective in single-turn settings, existing methods fail to maximize cumulative performance in multi-turn dialogue due to interaction dynamics and delayed rewards. To address this challenge, we move from myopic, single-turn selection to long-hor

Songping Peng, Zhiheng Zhang, Daojian Zeng, Lincheng Jiang

Safety alignment in Large Language Models (LLMs) remains highly fragile during fine-tuning, where even benign adaptation can degrade pre-trained refusal behaviors and enable harmful responses. Existing defenses typically constrain either weights or activations in isolation, without considering their coupled effects on safety. In this paper, we first theoreti

Chen Zhou, Jiangtao Luo, Yongyi Ran

Low Earth Orbit (LEO) satellite networks provide global coverage and low latency, yet high node mobility, uneven traffic distribution, and stochastic link failures pose severe challenges for inter-domain routing. Existing approaches either neglect graph-structured topology or lack dynamic awareness of real-time link states, struggling to balance load distrib

Han Liu, Jiaqi Li, Zhi Xu, Xiaotong Zhang

Black-box adversarial attack on vision-language pre-trained models is a practical and challenging task, as text and image perturbations need to be considered simultaneously, and only the predicted results are accessible. Research on this problem is in its infancy, and only a handful of methods are available. Nevertheless, existing methods either rely on a co

Hao Wang, Jiqing Zhang, Xin Yang, Baocai Yin

Camouflaged Object Detection (COD) aims to segment objects that blend seamlessly into complex backgrounds, with growing interest in exploiting additional visual modalities to enhance robustness through complementary information. However, most existing approaches generally rely on modality-specific architectures or customized fusion strategies, which limit sc

Yuangang Li, Justin Tian Jin Chen, Ethan Yu, David Hong

Large language models (LLMs) increasingly rely on explicit reasoning to solve coding tasks, yet evaluating the quality of this reasoning remains challenging. Existing reasoning evaluators are not designed for coding, and current benchmarks focus primarily on code generation, leaving other coding tasks largely unexplored. We introduce CodeRQ-Bench, the first

Daniil Gurgurov, Tom Röhr, Sebastian von Rohrscheidt, Josef van Genabith

Despite advances in multilingual capabilities, most large language models (LLMs) remain English-centric in their training and, crucially, in their production of reasoning traces. Even when tasked with non-English problems, these models predominantly reason in English, creating a fundamental mismatch for non-English usage scenarios. We address this disparity

SungHo Kim, Juhyeong Park, Eda Atalay, SangKeun Lee

Korean is a morphologically rich language with a featural writing system in which each character is systematically composed of subcharacter units known as Jamo. These subcharacters not only determine the visual structure of Korean but also encode frequent and linguistically meaningful morphophonological processes. However, most current Korean language models

S. R. He, L. Gao, P. H. Ouyang, H. Zheng

The resonant cavity system has been widely used to search for the electromagnetic response of dark photons, although its achievable detection sensitivity remains at a relatively low level. In this letter, we propose a feasible approach to significantly improve its achievable detection sensitivity by enhancing the detectability of the dark photon-photon dynam

Amilcare Porporato, Lamberto Rondoni

We revisit the Rayleigh--Riabouchinsky paradox in dimensional analysis by making explicit the bridge between thermodynamics and the mechanical interpretation of temperature. Boltzmann's constant $k_B$ acts as a dimensional unifier, leading to an augmented $\Pi$-theorem with an additional dimensionless group that encodes system size. In the macroscopic thermo

NVIDIA, :, Aakshita Chandiramani, Aaron Blakeman

We describe the pre-training, post-training, and quantization of Nemotron 3 Super, a 120 billion (active 12 billion) parameter hybrid Mamba-Attention Mixture-of-Experts model. Nemotron 3 Super is the first model in the Nemotron 3 family to 1) be pre-trained in NVFP4, 2) leverage LatentMoE, a new Mixture-of-Experts architecture that optimizes for both accurac

Tomer Ashuach, Shai Gretz, Yoav Katz, Yonatan Belinkov

Humans use introspection to evaluate their understanding through private internal states inaccessible to external observers. We investigate whether large language models possess similar privileged knowledge about answer correctness, information unavailable through external observation. We train correctness classifiers on question representations from both a

Ravikumar Balakrishnan, Sanket Mendapara, Ankit Garg

We study typographic prompt injection attacks on vision-language models (VLMs), where adversarial text is rendered as images to bypass safety mechanisms, posing a growing threat as VLMs serve as the perceptual backbone of autonomous agents, from browser automation and computer-use systems to camera-equipped embodied agents. In practice, the attack surface is

Gregory Morse, Tamás Kozsik

We study the problem of maintaining a breadth-first spanning tree and the induced BFS ordering in a directed graph under edge updates. While semi-dynamic algorithms are known, maintaining the spanning tree, level information, and numbering together in the fully dynamic setting is less developed. This preprint presents a framework for fully dynamic BFS in dir

Stephen Wiggins

Out-of-Time-Order Correlators (OTOCs) quantify quantum information scrambling, but their connection to localized phase-space structures, such as chemical transition states, requires formal development. We derive a leading-order semiclassical expansion for the local microcanonical OTOC in systems with an index-1 saddle point, expressing the scrambling rate as

Davit Gondauri

In the face of socioeconomic challenges, this paper develops and empirically demonstrates the Gondauri Index (GI) as a reproducible diagnostics-first composite framework for benchmarking macro-financial resilience across heterogeneous economies on a unified 0-100 scale. The GI addresses a key limitation of conventional surveillance dashboards: resilience is

Shuhui Liu, Riccardo Tomasello, Bin Fang, Aitian Chen

The development of compact and highly sensitive microwave detectors compatible with complementary-metal-oxide-semiconductor (CMOS) processes is an active research area but remains a major challenge in microwave technology. Spin-torque diodes (STDs) are emerging nanoscale spintronic devices capable of surpassing the theoretical thermodynamic sensitivity limit

Chenlin Zhou, Sihang Guo, Jiaqi Wang, Dongyang Ma

Regarded as the third generation of neural networks, Spiking Neural Networks (SNNs) have garnered significant traction due to their biological plausibility and energy efficiency. Recent advancements in large models necessitate spiking neurons capable of high performance, adaptability, and training efficiency. In this work, we first propose a novel functional

Gregor Krzmanc, Vinicius Mikuni, Benjamin Nachman, Callum Wilkinson

Future AI-based studies in particle physics will likely start from a foundation model to accelerate training and enhance sensitivity. As a step toward a general-purpose foundation model for particle physics, we investigate whether the OmniLearned and ParticleViT foundation models pretrained on diverse high-$Q^2$ simulated and real $pp$ and $ep$ collisions re

Shubhada Aute, Fahad Panolan, Geevarghese Philip

Motivated by the landmark resolution of the 1-2-3 Conjecture, we initiate the study of the parameterized complexity of the Vertex-Coloring {0,1}-Edge-Weighting problem and its generalization, Vertex-Coloring Pre-edge-Weighting, under various structural parameters. The base problem, Vertex-Coloring {0,1}-Edge-Weighting, asks whether we can assign a weight fro

Tatsuro Kawakami, Jakub Witaszek

We introduce higher $F$-rationality generalising $F$-rationality. We prove that a normal variety over a field of characteristic zero is $m$-rational if and only if it is $m$-$F$-rational after reduction modulo a sufficiently large prime $p$. Additionally, we establish new results on the logarithmic extension of forms.

Tanveer Ahmad

We present a comprehensive theoretical analysis of ultrafast entanglement generation between two Rydberg-blockaded atoms, explicitly accounting for realistic laser noise. Using femtosecond Gaussian pulses as a baseline, we systematically evaluate Bell-state fidelity sensitivity to amplitude and phase noise across white, pink (1/f), and Ornstein-Uhlenbeck spe

Hayata Matsuda, Ruo Hibino, Chihiro Tabata, Koji Kaneko

The spontaneous Hall effect, corresponding to a zero-field anomalous Hall effect (AHE), is induced by symmetry breaking associated with ferromagnetism. Studies in recent years, however, have revealed that antiferromagnetic (AFM) states characterized by magnetic point groups that allow ferromagnetism can also break the relevant symmetries and induce AHE witho

Rui Yin, Tianxu Han, Naen Xu, Changjiang Li

Safety-aligned large language models (LLMs) are increasingly deployed in real-world pipelines, yet this deployment also enlarges the supply-chain attack surface: adversaries can distribute backdoored checkpoints that behave normally under standard evaluation but jailbreak when a hidden trigger is present. Recent post-hoc weight-editing methods offer an effic

Jiwan Kim, Kibum Kim, Wonjoong Kim, Byung-Kwan Lee

Recently, visual token pruning has been studied to handle the vast number of visual tokens in Multimodal Large Language Models. However, we observe that while existing pruning methods perform reliably on simple visual understanding, they struggle to effectively generalize to complex visual reasoning tasks, a critical gap underexplored in previous studies. Th

Kyungmin Min, Minbeom Kim, Kang-il Lee, Seunghyun Yoon

Detailed image captioning demands both factual grounding and fine-grained coverage, yet existing methods have struggled to achieve them simultaneously. We address this tension with Reflective Note-Guided Captioning (ReflectCAP), where a multi-agent pipeline analyzes what the target large vision-language model (LVLM) consistently hallucinates and what it syst

Dongjian Yu, Weiqing Min, Qian Jiang, Xing Lin

Accurate estimation of food nutrition plays a vital role in promoting healthy dietary habits and personalized diet management. Most existing food datasets primarily focus on Western cuisines and lack sufficient coverage of Chinese dishes, which restricts accurate nutritional estimation for Chinese meals. Moreover, many state-of-the-art nutrition prediction m

Mikhailo Dokuchaev, Juan Jacobo Simón

In this paper, we extend the study of graded equivalences to the case of general idempotent graded rings. We prove that the existence of a graded equivalence between two categories of graded torsion-free unital modules may be characterized by the existence of a Morita context with surjective trace maps. As an application of our results we relate certain latt

Qing-Wei Wang

One intriguing property of non-Hermitian systems is the breakdown of adiabatic theorem and chiral state conversion as the system dynamically encircles exceptional points. However, the subtle dependence of the chiral dynamics on the loop geometry, the starting point, the encircling speed and especially the noise has not been studied systematically. Here we pr

Haifeng Zhang, Qinghui He, Xiuli Bi, Bo Liu

In recent years, the rapid development of generative artificial intelligence technology has significantly lowered the barrier to creating high-quality fake images, posing a serious challenge to information authenticity and credibility. Existing generated image detection methods typically enhance generalization through model architecture or network design. Ho

Joongmin Shin, Chanjun Park, Jeongbae Park, Jaehyung Seo

RAG-based QA has emerged as a powerful method for processing long industrial documents. However, conventional text chunking approaches often neglect complex and long industrial document structures, causing information loss and reduced answer quality. To address this, we introduce MultiDocFusion, a multimodal chunking pipeline that integrates: (i) detection o

Yi Xiong, Liang Xiong, Xiaohong Ji, Sen Yang

Molecular property optimization is central to drug discovery, yet many deep learning methods rely on black-box scoring and offer limited control over scaffold preservation, often producing unstable or biologically implausible edits. While large language models (LLMs) are promising molecular generators, optimization remains constrained by the lack of chemistr

Annalisa Degenhard, Sophia Ppali, Fotis Liarokapis, Enrico Rukzio

Virtual reality exposure therapy (VRET) enables controlled exposure to trauma-related stimuli to facilitate memory access and emotional processing. However, the field remains underexplored for complex post-traumatic stress disorder (C-PTSD). Unlike single-trauma PTSD, C-PTSD requires highly individualized triggers that are difficult to identify and implement

Parthaw Goswami, Md Khairul Islam, Ashfak Yeafi

Federated learning (FL) enables collaborative model training without sharing raw data, offering a promising path toward privacy preserving artificial intelligence. However, FL models may still memorize sensitive information from participants, conflicting with the right to be forgotten (RTBF). To meet these requirements, federated unlearning has emerged as a

Wuping Yang, H. Huang

Self-healing is the ability of a wave packet to spontaneously restore its spatial profile after scattering. As an emergent feature of non-unitary dynamics, it has attracted significant interest in non-Hermitian physics. Here, we systematically investigate how stochastic noise influences edge self-healing. Counterintuitively, we find that noise can constructi

Zanyi Wang, Fan Li, Dengyang Jiang, Liuzhuozheng Li

Spatio-temporal video grounding (STVG) aims to localize queried objects within dynamic video segments. Prevailing fully-trained approaches are notoriously data-hungry. However, gathering large-scale STVG data is exceptionally challenging: dense frame-level bounding boxes and complex temporal language alignments are prohibitively expensive to annotate, especi

Avadhut V. Purohit, Udaysinh T. Bhosale

The quantum kicked top (QKT) is one of the most widely studied models in quantum chaos, providing a minimal yet powerful framework for exploring the relationship between classical nonlinear dynamics and quantum behavior. Unlike many chaotic systems with infinite-dimensional Hilbert spaces, the QKT possesses a finite-dimensional Hilbert space, making it analy

Bin Zhang, Yabiao Wang, Xiaoyao Xie, Shanping You

The exponential growth of data from modern radio telescopes presents a significant challenge to traditional single-pulse search algorithms, which are computationally intensive and prone to high false-positive rates due to Radio Frequency Interference (RFI). In this work, we introduce FRTSearch, an end-to-end framework unifying the detection and physical char

Huy Anh Nguyen, Feras Dayoub, Minh Hoai

We address the challenging task of detecting the precise moment when hands make contact with objects in egocentric videos. This frame-level detection is crucial for augmented reality, human-computer interaction, assistive technologies, and robot learning applications, where contact onset signals action initiation or completion. Temporally precise detection i

Qi Li, Cheng-Long Wang, Yinzhi Cao, Di Wang

Training models on a carefully chosen portion of data rather than the full dataset is now a standard preprocess for modern ML. From vision coreset selection to large-scale filtering in language models, it enables scalability with minimal utility loss. A common intuition is that training on fewer samples should also reduce privacy risks. In this paper, we cha

Gilhan Kim

We decompose the Kullback--Leibler generalization error (GE) -- the expected KL divergence from the data distribution to the trained model -- of unsupervised learning into three non-negative components: model error, data bias, and variance. The decomposition is exact for any e-flat model class and follows from two identities of information geometry: the gene

Si-Qi Du, Guo-Zhu Song, Hai-Rui Wei

High-dimensional quantum systems offer a number of advantages in larger information capacity, stronger noise resiliency, higher improved efficiency and accuracy over the qubit systems. In quantum communication the maximally entangled states will inevitably become mixed states or less-entangled pure states by the channel noise during the practical transmissio

Charlotte S. Alexander, Shane Storks, Souradip Pal, Sayak Chakrabarty

Letters of recommendation (LoRs) can carry patterns of implicitly gendered language that can inadvertently influence downstream decisions, e.g. in hiring and admissions. In this work, we investigate the extent to which Transformer-based encoder models as well as Large Language Models (LLMs) can infer the gender of applicants in academic LoRs submitted to an

Yue Wu, Yuan-Ting Zhong, Ze-Yuan Ma, Yue-Jiao Gong

Streaming Data-Driven Optimization (SDDO) problems arise in many applications where data arrive continuously and the optimization environment evolves over time. Concept drift produces non-stationary landscapes, making optimization methods challenging due to outdated models. Existing approaches often rely on simple surrogate combinations or directly injecting

Tanzila Rahman, Renjie Liao, Leonid Sigal

Training multimodal large language models (MLLMs) for video understanding requires large-scale annotated data spanning diverse tasks such as object counting, question answering, and segmentation. However, collecting and annotating multimodal video data in real-world is costly, slow, and inherently limited in diversity and coverage. To address this challenge,

Hao Wu

Let $X$ be a compact Riemann surface and let $L$ be a positive line bundle on $X$. We obtain the growth speed of unit ball volume in $H^0(X,L^n)$ towards the energy at equilibrium. As an application, we also obtain the speed of Fekete measures converging to the equilibrium measure.

Ziyang Zhao, Haoran Xiong, Zicheng Ye, Luyi Li

LDPC codes have attracted significant attention due to their capacity-approaching performance. Elementary trapping sets are the main cause of the error floor phenomenon in LDPC codes. We investigate several graph structures associated with trapping sets, including theta graphs, dumbbell graphs, and short cycles with chords. Based on the Turán numbers of $θ(2

Ivannia Gomez Moreno, Yi Yao, Ye Tian, Xiaofan Yu

LiDAR semantic segmentation plays a pivotal role in 3D scene understanding for edge applications such as autonomous driving. However, significant challenges remain for real-world deployments, particularly for on-device post-deployment adaptation. Real-world environments can shift as the system navigates through different locations, leading to substantial per

Ned Goodman, Alexander S. Dellios, Margaret D. Reid, Peter D. Drummond

Quantum computers solve intractable problems which classically require an exponentially long time to compute. With the development of large-scale experiments that claim quantum advantage, a vital issue has now emerged. What are the errors, and how do they affect the complexity of the problem solved? Large-scale Gaussian boson sampling (GBS) experiments give

Shuyi Miao, Wangjie Qiu, Shengda Zhuo, Fei Shen

As cross-chain interoperability advances, decentralized finance (DeFi) protocols enable illicit funds to be reorganized into uniform liquid assets that flow throughout the cryptocurrency market. Such operations can bypass monitoring targeted at individual blockchains and thereby weaken current regulatory frameworks. Motivated by these, we introduce UniDetect

Chu-Tian Gao, Chen Lu, Yu-Bo Liu, Zhiming Pan

Unconventional superconductivity (SC) characterized by multi-fermion orderings has attracted substantial attention. However, previous studies have largely focused on 2D systems or 3D systems with effective 2D symmetries. Here, we investigate the vestigial phases arising from thermal fluctuations of chiral SC in 3D systems governed by the cubic $O_h$ point gr

Marieke Stolte, Jörg Rahnenführer, Andrea Bommert

Methods for quantifying the similarity of datasets are relevant in applications where two or more datasets, or their underlying distributions, need to be compared, ranging from two- and k-sample testing to applications in machine learning and synthetic data generation. Many methods for quantifying the similarity of datasets are available from the literature,

Iori Ishiguro, Hayate Kunitsu, Natsuki Mitsuishi, Shunsuke Tsuda

We investigated the energy gap formation across the charge density wave (CDW) transition inof ZrTe_{2.98}Se_{0.02}. By employing a laser photoemission microscopy, we clearly resolved one elliptical Fermi surface (FS) around the Brillouin zone (BZ) center, and two quasi-one-dimensional FSs along the BZ boundary. We further mapped the intensity difference betw

Nivedita Batra, Chiranjoy Chattopadhyay, Mayurakshi Chaudhuri

Reliable analysis of migration is critically dependent on the quality and consistency of the underlying data. Indian migration data, primarily derived from decennial census records, are affected by systematic gaps arising from uneven coverage and measurement inconsistencies across states and time. This paper presents a data-centric framework, HICM, for harmo

Deyuan Liu, Peng Sun, Yansen Han, Zhenglin Cheng

The push for efficient text to image synthesis has moved the field toward one step sampling, yet existing methods still face a three way tradeoff among fidelity, inference speed, and training efficiency. Approaches that rely on external discriminators can sharpen one step performance, but they often introduce training instability, high GPU memory overhead, a

Boyang Li, Hongzhe Shou, Yuanyuan Liang, Jingbin Zhang

Existing Chinese toxic content detection methods mainly target sentence-level classification but often fail to provide readable and contiguous toxic evidence spans. We propose \textbf{ToxiTrace}, an explainability-oriented method for BERT-style encoders with three components: (1) \textbf{CuSA}, which refines encoder-derived saliency cues into fine-grained to

Jianzhe Ma, Zhonghao Cao, Shangkui Chen, Yichen Xu

While video large language models (Video-LLMs) excel in understanding slow-paced, real-world egocentric videos, their capabilities in high-velocity, information-dense virtual environments remain under-explored. Existing benchmarks focus on daily activities, yet lack a rigorous testbed for evaluating fast, rule-bound reasoning in virtual scenarios. To fill th

Guoan Xu, Yang Xiao, Guangwei Gao, Dongchen Zhu

Multimodal semantic segmentation has emerged as a powerful paradigm for enhancing scene understanding by leveraging complementary information from multiple sensing modalities (e.g., RGB, depth, and thermal). However, existing cross-modal fusion methods often implicitly assume that all modalities are equally reliable, which can lead to feature degradation whe

Hayato Inoue, Shota Harada, Shumpei Takezaki, Ryoma Bise

Existing cell instance segmentation pipelines typically combine deterministic predictions with post-processing, which imposes limited explicit constraints on the global structure of instance masks. In this work, we propose a multi-task image-to-image Schr\"odinger Bridge framework that formulates instance segmentation as a distribution-based image-to-image g

Mingkun Ye

In this paper, we establish the existence of weak solutions for distribution-dependent stochastic differential equations (DDSDEs) driven by a broad class of L\'{e}vy noises, where the drift coefficients satisfy specific integrability conditions. This is achieved through the Krylov-type estimate and tightness argument.

Giuliano Benenti, Giulio Casati, Jiangbin Gong, Zhixing Zou

The kicked rotor provides a simple yet powerful model for introducing many of the central concepts of classical and quantum chaos. Despite its apparent simplicity, it exhibits rich dynamical behavior and has found applications across a wide range of fields, including atomic and optical physics, condensed matter physics, and emerging quantum technologies. Thi

Zhiwei Zhang, Xingyuan Zeng, Xinkai Kong, Kunquan Zhang

Agricultural parcel extraction plays an important role in remote sensing-based agricultural monitoring, supporting parcel surveying, precision management, and ecological assessment. However, existing public benchmarks mainly focus on regular and relatively flat farmland scenes. In contrast, terraced parcels in mountainous regions exhibit stepped terrain, pro

Bang Quan Zheng

Racial differences in authoritarianism are widely used to explain variation in political attitudes, yet it is unclear whether they reflect true latent differences or measurement artifacts. Using anchor-based multi-group confirmatory factor analysis across multiple nationally representative surveys, this paper examines measurement equivalence in the standard

Jingbo Yang, Guanyu Yao, Bairu Hou, Xinghan Yang

As Large Language Models (LLMs) are increasingly deployed as task-oriented agents in enterprise environments, ensuring their strict adherence to complex, domain-specific operational guidelines is critical. While utilizing an LLM-as-a-Judge is a promising solution for scalable evaluation, the reliability of these judges in detecting specific policy violations

S M Jamil Uddin

The emergence of vibe coding, a paradigm where non-technical users instruct Large Language Models (LLMs) to generate executable codes via natural language, presents both significant opportunities and severe risks for the construction industry. While empowering construction personnel such as the safety managers, foremen, and workers to develop tools and softw