October 2025 arXiv papers — page 114

Janne Rotter, William Bailkoski

AI has the potential to significantly improve how NGOs utilize their limited resources for societal benefits, but evidence about how NGOs adopt AI remains scattered. In this study, we systematically investigate the types of AI adoption use cases in NGOs and identify common challenges and solutions, contextualized by organizational size and geographic context

Naoki Yoshida, Satoshi Hayakawa, Yuhta Takida, Toshimitsu Uesaka

In this study, we propose an enhancement to the similarity computation mechanism in multi-modal contrastive pretraining frameworks such as CLIP. Prior theoretical research has demonstrated that the optimal similarity metrics between paired modalities should correspond to the pointwise mutual information (PMI) between the two modalities. However, the current

Stanisław Sieniawski, Rafał Demkowicz-Dobrzański

We present an efficient tensor-network based algorithm for finding the optimal adaptive quantum channel discrimination strategies inspired by recently developed numerical methods in quantum metrology to find the optimal adaptive channel estimation protocols. We examine the connection between channel discrimination and estimation problems, highlighting in par

Aron Distelzweig, Faris Janjoš, Oliver Scheel, Sirish Reddy Varra

Traditionally, prediction and planning in autonomous driving (AD) have been treated as separate, sequential modules. Recently, there has been a growing shift towards tighter integration of these components, known as Integrated Prediction and Planning (IPP), with the aim of enabling more informed and adaptive decision-making. However, it remains unclear to wh

L. Pigatto, G. Frello, Y. Q. Liu, L. Novello

JT-60SA is a large superconducting tokamak built in Naka, Japan. After the successful achievement of its first MA-class plasma, the installation of several additional sub-systems, including a set of non-axisymmetric Error Field Correction Coils (EFCC), is ongoing. Optimization of future JT-60SA plasma scenarios will critically depend on the correct use of EF

Shijia Kang, Muhan Zhang

Reinforcement learning (RL) has been pivotal in enhancing the reasoning capabilities of large language models (LLMs), but it often suffers from limited exploration and entropy collapse, where models exploit a narrow set of solutions, leading to a loss of sampling diversity and subsequently preventing RL from further improving performance. This issue is exace

Yao Huang, Yitong Sun, Yichi Zhang, Ruochen Zhang

Despite the remarkable advances of Large Language Models (LLMs) across diverse cognitive tasks, the rapid enhancement of these capabilities also introduces emergent deceptive behaviors that may induce severe risks in high-stakes deployments. More critically, the characterization of deception across realistic real-world scenarios remains underexplored. To bri

Johann Ostmeyer, Carsten Urbach

We introduce a new method to approximate Euclidean correlation functions by exponential sums. The Truncated Hankel Correlator (THC) method builds a Hankel matrix from the full correlator data available and truncates the eigenspectrum of said Hankel matrix. It proceeds by applying the Prony generalised eigenvalue method to the thus obtained low-rank approxima

Asaki Saito, Jun-Ichi Tamura

The objective of this paper is to show (a)=(b)=(c) as rational functions of $T$, $U$ for (a), (b), (c) given by (a) continued fractions of length $2^{n+1}-1$ with explicit partial denominators in $\left\{-T,U^{-1}T\right\}$, (b) truncated series $\sum_{0\le m\le n} \left(U^{2^m}/\left(h_0(T)h_1(T,U) \cdots h_m(T,U)\right)\right)$ with $h_n$ defined by $h_0:=

Xianmin Chen, Peiliang Huang, Longfei Han, Dingwen Zhang

With the rapid development of deep learning, low-light RAW image enhancement (LLRIE) has achieved remarkable progress. However, the challenge that how to simultaneously achieve strong enhancement quality and high efficiency still remains. Leveraging the inherent efficiency of Channel Attention and Mamba, we introduce a Hierarchical Mixing Architecture (HiMA)

Jade Leathrum

We investigate modified Sierpiński Carpet fractals, constructed by dividing a square into a square $n \times n$ grid, removing a subset of the squares at each step, and then repeating that process for each square remaining in that grid. If enough squares are removed and in the proper places, we get ``Dust Type'' carpets, which have a path-connected c

Woo-Jin Ahn, Sang-Ryul Baek, Yong-Jun Lee, Hyun-Duck Choi

Reinforcement learning algorithms typically utilize an interactive simulator (i.e., environment) with a predefined reward function for policy training. Developing such simulators and manually defining reward functions, however, is often time-consuming and labor-intensive. To address this, we propose an Offline Simulator (OffSim), a novel model-based offline

Lirong Yi, Gregory Gay, Philipp Leitner

Large Language Models (LLMs) can generate code, but can they generate fast code for complex, real-world software systems? In this study, we investigate this question using a dataset of 65 tasks mined from performance-critical open-source Java projects. Unlike prior studies, which focused on algorithmic puzzles, we conduct experiments on actual performance-se

Yui Hayashi, Masashi Kawahira, Hiromasa Watanabe

We study the finite-temperature phase structure of the four-dimensional ${\rm SU}(2)$ adjoint Higgs model, focusing on a possible \textit{deconfinement-Higgs continuity}: the conjecture that the high-temperature deconfined phase of Yang-Mills theory and the finite-temperature Higgs phase form a single thermodynamic phase. We first perform a global-symmetry a

Maxence Lefèvre, Matteo Cerminara, Antonio Costa

Volcanism on Venus has never been directly observed, but several measurements indicate present-day activity. Volcanism could potentially play a role in climatic processes on Venus, especially in the sulfur cycle like on Earth. Observation of volcanic activity is the primary objective of future Venus spacecraft. However, there are many unknowns regarding its

Andre Rochow, Jonas Marcic, Svetlana Seliunina, Sven Behnke

3D phenotyping of plants plays a crucial role for understanding plant growth, yield prediction, and disease control. We present a pipeline capable of generating high-quality 3D reconstructions of individual agricultural plants. To acquire data, a small commercially available UAV captures images of a selected plant. Apart from placing ArUco markers, the entir

Diogo Landau, Gijs Blanken, Jorge Barbosa, Nishant Saurabh

Modern distributed systems rely on complex networks of interconnected services, creating direct or indirect dependencies that can propagate faults and cause cascading failures. To localize the root cause of performance degradation in these environments, constructing a service dependency graph is highly beneficial. However, building an accurate service depend

Manit Mishra

Purpose: The purpose of this study is to map the body of scholarly literature at the intersection of artificial intelligence (AI), analytics and sports and thereafter, leverage the insights generated to chart guideposts for future research. Design/methodology/approach: The study carries out systematic literature review (SLR). Preferred Reporting Items for Sy

Sergio Muñiz Subiñas, Manuel L. González, Jorge Ruiz Gómez, Alejandro Mata Ali

This work introduces a post-training quantization (PTQ) method for dense neural networks via a novel ADAROUND-based QUBO formulation. Using the Frobenius distance between the theoretical output and the dequantized output (before the activation function) as the objective, an explicit QUBO whose binary variables represent the rounding choice for each weight an

Kerem Bükrü, Steffen Leger, M. Lautaro Hickmann, Hans-Martin Rieser

Gaussian processes are widely known for their ability to provide probabilistic predictions in supervised machine learning models. Their non-parametric nature and flexibility make them particularly effective for regression tasks. However, training a Gaussian process model using standard methods requires matrix inversions with a cubic time complexity, which po

Jing He, Hua Jiang, Cheng Li, Siqian Xin

This work, based on Random Matrix Theory (RMT), introduces a novel early-stopping strategy for Transformer training dynamics. Utilizing the Power Law (PL) fit to tansformer attention matrices as a probe, we demarcate training into three stages: structural exploration, heavy-tailed structure stabilization, and convergence saturation. Empirically, we observe t

Dāvis Kažemaks, Laurens Versluis, Burcu Kulahcioglu Ozkan, Jérémie Decouchant

Apache Spark is a widely adopted framework for large-scale data processing. However, in industrial analytics environments, Spark's built-in schedulers, such as FIFO and fair scheduling, struggle to maintain both user-level fairness and low mean response time, particularly in long-running shared applications. Existing solutions typically focus on job-level fa

Krzysztof Sośnica, Agnès Fienga, Dmitry Pavlov, Nicolas Rambaux

All future lunar missions require a definition of the lunar reference system and a realization in the form of the lunar reference frame to ensure consistent products for positioning, navigation, cartography, and timing. This paper defines the origin, orientation, and scale of the Lunar Reference System (LRS), as well as provides numerical solutions for the f

Samuel Girard, Aurelien Bibaut, Arthur Gretton, Nathan Kallus

We study the problem of stochastic contextual bandits in the agnostic setting, where the goal is to compete with the best policy in a given class without assuming realizability or imposing model restrictions on losses or rewards. In this work, we establish the first fast rate for regret relative to the best-in-class policy. Our proposed algorithm updates the

Terry Mart, Jovan Alfian Djaja

A new elementary operator for kaon photoproduction on the nucleon and nuclei has been developed within a Feynman diagrammatic framework. By fitting the unknown coupling strengths at the electromagnetic and hadronic vertices of the baryon resonances to all available experimental data across the six isospin channels, the model achieves excellent agreement with

Carlos Arranz-Simón, Begoña Cano, César Palencia

Given an $A$-stable rational approximation to $e^z$ of order $p$, numerical procedures are suggested to time integrate abstract, well-posed IBVPs, with time-dependent source term $f$ and boundary value $g$. These procedures exhibit the optimal order $p$ and can be implemented by using just one single evaluation of $f$ and $g$ per step, i.e., no evaluations o

Muslim Chochlov, Gul Aftab Ahmed, James Vincent Patten, Yuanhua Han

Source code clones pose risks ranging from intellectual property violations to unintended vulnerabilities. Effective and efficient scalable clone detection, especially for diverged clones, remains challenging. Large language models (LLMs) have recently been applied to clone detection tasks. However, the rapid emergence of LLMs raises questions about optimal

Shiqin Tang, Rong Feng, Shuxin Zhuang, Youzhi Zhang

We study counterfactual prediction under assignment bias and propose a mathematically grounded, information-theoretic approach that removes treatment-covariate dependence without adversarial training. Starting from a bound that links the counterfactual-factual risk gap to mutual information, we learn a stochastic representation Z that is predictive of outcom

A. Golebiewska, S. Rybicki, P. Stefaniak

The aim of this paper is to formulate necessary conditions and sufficient ones for the existence of closed connected sets of nonstationary $2 \pi$-periodic solutions of $S^1$-symmetric Newtonian systems in $C_{2 \pi}([0,2\pi],\Omega) \times (0,+ \infty)$. As the main topological tool we apply the degree for equivariant gradient maps.

Marc Benjamin Hahn, Sepideh Aminzadeh-Gohari, Anna Grebinyk, Matthias Gross

Cancer treatment with radiotherapy aims to kill tumor cells and spare healthy tissue.Thus,the experimentally observed sparing of healthy tissue by the FLASH effect during irradiations with ultra-high dose rates (UHDR) enables clinicians to extend the therapeutic window.However, the underlying radiobiological and chemical mechanisms are far from being underst

Maxence Lefèvre, Sébastien Lebonnois, Aymeric Spiga, François Forget

The knowledge of the Venus near-surface atmosphere is sparse. Few spacecrafts landed on the surface and measured winds with amplitudes below 1 m/s. The diurnal cycle of the wind amplitude and orientation is not known. Recent numerical simulations showed that slope winds along topographic structures could strongly impact the direction of winds. This study pre

M. H. Duong, M. J. Reynolds

The purpose of this paper is to propose a revised continuum model from the discrete system introduced in [Deng et.al., PRL, 2017] . Using a Galilean transformation, we obtain an equation governing the soliton solutions in the phase plane - a second-order nonlinear ODE related to the Klein-Gordon equation with quadratic nonlinearity. These admit the well-know

Mathieu Benninghoff, Gilles Vilmart

Explicit stabilized methods are highly efficient time integrators for large and stiff systems of ordinary differential equations especially when applied to semi-discrete parabolic problems. However, when local spatial mesh refinement is introduced, their efficiency decreases, since the stiffness is driven by only the smallest mesh element. A natural approach

Ruoke Meng, Geert Smet, Dieter Van den Bleeken, Aaron Van Poecke

Evaluations are presented for the prediction of wind power ramping events in the Belgian Offshore Zone. Two models from the Royal Meteorological Institute of Belgium are verified: the operational ALARO-4km and its version with Wind Farm Parameterization (WFP). Power predictions are produced using power curves and machine learning (ML). As standard metrics su

Petra Berenbrink, Robert Elsässer, Tom Friedetzky, Hamed Hosseinpour

We consider discrete, iterative load balancing via matchings on arbitrary graphs. Initially each node holds a certain number of tokens, defining the load of the node, and the objective is to redistribute the tokens such that eventually each node has approximately the same number of tokens. We present results for a general class of simple local balancing sche

Dawei Dai, Yinxiu Zhou, Chenghang Li, Guolai Jiang

In facial image generation, current text-to-image models often suffer from facial attribute leakage and insufficient physical consistency when responding to local semantic instructions. In this study, we propose Face-MakeUpV2, a facial image generation model that aims to maintain the consistency of face ID and physical characteristics with the reference imag

Chloé Van Bastelaere, Felix A. Palm, Botao Wang, Nathan Goldman

This work investigates the coexistence of distinct topologically ordered phases within a single setup. We demonstrate this concept through tensor network simulations of the Hofstadter-Bose-Hubbard model under a spatially modulated chemical potential. Focusing on cylindrical geometries, we realize regions exhibiting the Laughlin-1/2 phase and its particle-hol

Vu Tram Anh Khuong, Thi Bich Phuong Man, Luu Tu Nguyen, Thanh Ha Le

Facial micro-expressions are brief, involuntary facial movements that reveal hidden emotions. Most Micro-Expression Recognition (MER) methods that rely on optical flow typically focus on the onset-to-apex phase, neglecting the apex-to-offset phase, which holds key temporal dynamics. This study introduces a Combined Optical Flow (COF), integrating both phases

J O Button

We examine second bounded cohomology and mod p homology in finite index subgroups of 1-relator groups and groups with a presentation of deficiency at least one. We use this to determine exactly which 1-relator groups are boundedly generated, as well as the groups of deficiency at least one up to a class of groups that conjecturally do not exist.

Shupei Lin, Nanfang Jiao, Yevhenii Shaidiuk, Delong Feng

We introduce and experimentally implement Fourier-plane phase synchronization for optical microscopy, and demonstrate its performance with interferometric scattering microscopy. By combining a photothermal phase plate and laser beam scanning, we realize a synchronized phase for all scattering components on the Fourier plane of high numerical-aperture microsc

Lingfeng Xuan, Chang Nie, Yiqing Xu, Zhe Liu

Structure from Motion (SfM) estimates camera poses and reconstructs point clouds, forming a foundation for various tasks. However, applying SfM to driving scenes captured by multi-camera systems presents significant difficulties, including unreliable pose estimation, excessive outliers in road surface reconstruction, and low reconstruction efficiency. To add

Vu Tram Anh Khuong, Luu Tu Nguyen, Thanh Ha Le, Thi Duyen Ngo

Micro-expressions (MEs) are brief, involuntary facial movements that reveal genuine emotions, typically lasting less than half a second. Recognizing these subtle expressions is critical for applications in psychology, security, and behavioral analysis. Although deep learning has enabled significant advances in micro-expression recognition (MER), its effectiv

Alicja Wierzcholska, Hubert Siejkowski

X-ray observations are essential to achieve a deeper understanding of the broadband emission mechanism in blazars. Here, we present a long-term spectral and temporal analysis of X-ray and optical observations of 1E 0229+200 collected with the Neil Gehrels Swift Observatory from 2008 to 2024, complemented by hard X-ray observations from the Nuclear Spectrosco

Nirmit Joshi, Gene Li, Siddharth Bhandari, Shiva Prasad Kasiviswanathan

We study the problem of learning to generate an answer (or completion) to a question (or prompt), where there could be multiple correct answers, any one of which is acceptable at test time. Learning is based on demonstrations of some correct answer to each training question, as in Supervised Fine Tuning (SFT). We formalize the problem as imitation learning (

Otto A. Hannuksela, K. Haris, Justin Janquart, Harsh Narola

Like light, gravitational waves are gravitationally lensed by intervening massive astrophysical objects, such as galaxies, clusters, black holes, and stars, resulting in a variety of potentially observable gravitational-wave lensing signatures. Searches for gravitational-wave lensing by the LIGO-Virgo-KAGRA (LVK) collaboration have begun. One common method f

Eddy Godelle

We prove that the lower central series of the cactus group associated with a non commutative Coxeter group never stabilizes. We also compute a minimal presentation in terms of generators for the cactus group associated with a finite Coxeter groups, except in type E.

Ginés López-Pérez, Esteban Martínez Vañó, Abraham Rueda Zoca

We show that there exists a Banach space in which every non-empty weakly open subset of its unit ball has radius one, the maximum possible value, but the infimum of the diameter of its slices is exactly one, so extremely far from its maximum. In fact, we show that there is a wide class of non-isomorphic Banach spaces satisfying this extreme difference betwee

Marco Merchand

We study inflation driven by a scalar potential arising from composite-sector dynamics, inspired by generalized composite Higgs models. The introduction of a non-minimal coupling, possessing the same functional form as the potential, induces a flattening at large field values that enables successful inflation. We analyze the conditions for ultra-slow-roll in

Tianyu Yang, Kangda Zhi, Shuangyang Li, Giuseppe Caire

In this work, we address the near-field imaging under a wideband wireless communication network by exploiting both the near-field channel of a uniform linear array (ULA) and the image correlation in the frequency domain. We first formulate the image recovery as a special multiple measurement vector (MMV) compressed sensing (CS) problem, where at various freq

Gabriele Visentin, Patrick Cheridito

In this paper, we show that interventionally robust optimization problems in causal models are continuous under the $G$-causal Wasserstein distance, but may be discontinuous under the standard Wasserstein distance. This highlights the importance of using generative models that respect the causal structure when augmenting data for such tasks. To this end, we

Chunhui Li, Chengrui Wang, Zhiqiang Yuan, Wei Fan

Comprehensive evaluation of the functionalities, algorithms, hardware components, and performance characteristics of future integrated sensing and communication (ISAC) base stations (BSs) under realistic deployment scenarios in controlled laboratory environments represents a critical requirement for ISAC technology advancement. A primary challenge in achievi

Jan Corazza, Hadi Partovi Aria, Daniel Neider, Zhe Xu

Reinforcement learning (RL) algorithms struggle with learning optimal policies for tasks where reward feedback is sparse and depends on a complex sequence of events in the environment. Probabilistic reward machines (PRMs) are finite-state formalisms that can capture temporal dependencies in the reward signal, along with nondeterministic task outcomes. While

Gucongcong Fan, Chaoyue Niu, Chengfei Lyu, Fan Wu

Mobile agents rely on Large Language Models (LLMs) to plan and execute tasks on smartphone user interfaces (UIs). While cloud-based LLMs achieve high task accuracy, they require uploading the full UI state at every step, exposing unnecessary and often irrelevant information. In contrast, local LLMs avoid UI uploads but suffer from limited capacity, resulting

Farjana Yesmin

Machine learning models trained on imbalanced datasets often exhibit intersectional biases-systematic errors arising from the interaction of multiple attributes such as object class and environmental conditions. This paper presents a data-driven framework for analyzing and mitigating such biases in image classification. We introduce the Intersectional Fairne

Rodrigo Guedes Lang, Tim Unbehaun, Lars Mohrmann, Simon Steinmassl

Imaging atmospheric Cherenkov telescopes (IACTs) are the main technique for detecting gamma rays with energies between tens of GeV and hundreds of TeV. Amongst them, the High Energy Stereoscopic System (H.E.S.S.) has pioneered the use of different telescope types to achieve an energy range as broad as possible. A large, 28 m diameter telescope is used in mon

Julien Poyatos, Octavi Fors, José Maria Gómez Cama

Stellar flares are intense bursts of radiation caused by magnetic reconnection on active stars. They are especially frequent on M dwarfs, where they can significantly influence the habitability of orbiting planets. Flare frequency distributions (FFDs) are typically modelled as power laws. However, recent studies challenge this assumption and propose alternat

Xiang Zhang, Michail Palaiologos, Christian Bluemm, Giuseppe Caire

IEEE 802.11ax introduces orthogonal frequency division multiple access (OFDMA) to WiFi to support concurrent transmissions to a larger number of users. As bandwidth continues to grow, WiFi channels exhibit increased frequency selectivity, which poses new challenges for MU-MIMO user selection: the optimal user set varies across frequency and is interleaved ov

Yu Luo, Shi-Gang Fang, Peng-Fa Li, Yuhao Feng

This paper proposes a wideband dual-polarized phased array with ultra-wideband scattering cross section (SCS) reduction. The antenna elements are loaded on a bilateral stepped ground. This ground is carefully designed in terms of height difference, step number, and length to achieve phase cancellation near the normal direction. Wideband dipoles with vertical

Albert Artiles

We prove that the return map of the unstable horocycle flow on the space of horizontally short translation surfaces associated to a lattice surface $(X, \omega)$ is weakly mixing. This extends a result of Cheung-Quas for the square torus to all lattice surfaces. The proof adapts their criterion for weakly mixing and uses quantitative bounds for Siegel-Veech

Zhiqiang Zhu, Xinbo Gao, Wen Lu, Jie Li

Existing nighttime aerial trackers based on prompt learning rely solely on spatial localization supervision, which fails to provide fine-grained cues that point to target features and inevitably produces vague prompts. This limitation impairs the tracker's ability to accurately focus on the object features and results in trackers still performing poorly. To

Nengbo Zhang, Hann Woei Ho

Recognizing the motion of Micro Aerial Vehicles (MAVs) is crucial for enabling cooperative perception and control in autonomous aerial swarms. Yet, vision-based recognition models relying only on RGB data often fail to capture the complex spatial temporal characteristics of MAV motion, which limits their ability to distinguish different actions. To overcome

Shiqin Tang, Shuxin Zhuang, Rong Feng, Runsheng Yu

Latent-variable energy-based models (LVEBMs) assign a single normalized energy to joint pairs of observed data and latent variables, offering expressive generative modeling while capturing hidden structure. We recast maximum-likelihood training as a saddle problem over distributions on the latent and joint manifolds and view the inner updates as coupled Wass

Ziang Guo, Zufeng Zhang

In autonomous driving, dynamic environment and corner cases pose significant challenges to the robustness of ego vehicle's state understanding and decision making. We introduce VDRive, a novel pipeline for end-to-end autonomous driving that explicitly models state-action mapping to address these challenges, enabling interpretable and robust decision making.

Qinxuan Wang, Chuang Wang, Mingyu Zhang, Jingwei Sun

Neural operators have emerged as a powerful data-driven paradigm for solving partial differential equations (PDEs), while their accuracy and scalability are still limited, particularly on irregular domains where fluid flows exhibit rich multiscale structures. In this work, we introduce the Multiscale Neural Operator (MNO), a new architecture for computationa

Gregory, Weintraub

Cloud data lakes provide a modern solution for managing large volumes of data. The fundamental principle behind these systems is the separation of compute and storage layers. In this architecture, inexpensive cloud storage is utilized for data storage, while compute engines are employed to perform analytics on this data in an "on-demand" mode. However, to ex

Zhi Zhou, Yuhao Tan, Zenan Li, Yuan Yao

Test-time scaling seeks to improve the reasoning performance of large language models (LLMs) by adding computational resources. A prevalent approach within the field is sampling-based test-time scaling methods, which enhance reasoning by generating multiple reasoning paths for a given input during inference. However, despite its practical success, the theore

Changda Zheng, Renshu Yang, Chenxi Ding, Songlin He

Cutting is a key factor affecting the speed of blasting excavation. With the continuous advancement of deep-hole blasting technology, determining the optimal extra-depth of the cut relative to the non-cut blast hole is of paramount importance. By combining model experiments and numerical simulations, this study systematically investigates the effect of extra

Roger Waldeck, Ann-Kristin Winkens, Clara Lemke, Carmen Leicht-Scholten

This workshop introduces participants to SUCRE, a serious game designed to enhance curriculum resilience in higher education by simulating crisis scenarios. While applicable to various disciplines, this session focuses on engineering curricula, identifying discipline-specific challenges and potential adaptations. Participants will engage in Step 1 of the gam

I. Sultana, A. Estrade, B. S. Meyer, H. Schatz

Type I X-ray bursts (XRBs) are thermonuclear runaways on the surface of accreting neutron stars, powered by rapid proton-capture and alpha-capture processes on neutron-deficient nuclei. Uncertainties in the corresponding reaction rates remain a major limitation in modeling burst light curves and ashes. We present a systematic study of the sensitivity of XRB

Xuchen Li, Xuzhao Li, Shiyu Hu, Kaiqi Huang

Long-form video reasoning remains a major challenge for Video Large Language Models (Video LLMs), as static uniform frame sampling leads to information dilution and obscures critical evidence. Furthermore, existing pixel-space video reasoning agents, which are designed to actively interact with the video to acquire new visual information, remain suboptimal d

Feifei Zhang, Zhenhong Jia, Sensen Song, Fei Shi

Despite the remarkable success of the end-to-end paradigm in deep learning, it often suffers from slow convergence and heavy reliance on large-scale datasets, which fundamentally limits its efficiency and applicability in data-scarce domains such as medical imaging. In this work, we introduce the Predictive-Corrective (PC) paradigm, a framework that decouple

Vincent Knight, Owen Campbell, Marc Harper, T. J. Gaffney

Direct reciprocity, typically studied using the Iterated Prisoner's Dilemma (IPD), is central to understanding how cooperation evolves. In the 1980s, Robert Axelrod organized two influential IPD computer tournaments, where Tit for Tat (TFT) emerged as the winner. Yet the archival record is incomplete: for the first tournament only a report survives, and for

Tongtao Ling, Shulin He, Pengjie Shen, Zhong-Qiu Wang

Multi-channel target speaker extraction (MC-TSE) aims to extract a target speaker's voice from multi-speaker signals captured by multiple microphones. Existing methods often rely on auxiliary clues such as direction-of-arrival (DOA) or speaker embeddings. However, DOA-based approaches depend on explicit direction estimation and are sensitive to microphone ar

Xiangchen Song, Yuchen Liu, Yaxuan Luan, Jinxu Guo

This study presents a controllable abstract summary generation method for large language models based on prompt engineering. To address the issues of summary quality and controllability in traditional methods, we design a multi-stage prompt generation framework. This framework generates summaries with varying levels of abstraction by performing semantic anal

Luo Long, Coralia Cartis, Paz Fink Shustin

Bayesian optimisation (BO) is a standard approach for sample-efficient global optimisation of expensive black-box functions, yet its scalability to high dimensions remains challenging. Here, we investigate nonlinear dimensionality reduction techniques that reduce the problem to a sequence of low-dimensional Latent-Space BO (LSBO). While early LSBO methods us

Huan Chen, Ting Han, Siyu Chen, Zhihao Guo

Street-view imagery (SVI) offers a fine-grained lens on traffic risk, yet two fundamental challenges persist: (1) how to construct street-level indicators that capture accident-related features, and (2) how to quantify their causal impacts across different accident types. To address these challenges, we propose Semantic4Safety, a framework that applies zero-

O. Heczko, F. Maca, V. Drchal, L. Fekete

Thermally induced antiphase boundaries (APBs) in ferromagnetic, ordered Ni-Mn-Ga single crystal exhibit complex, irregular shapes and closed loops without any lattice plane preferences. The APBs were visualized on polished (100) surface using magnetic force microscopy (MFM) at the same location in parent cubic austenite and monoclinic martensite with uniaxia

Kevin Wilkinghoff, Alessia Cornaggia-Urrigshardt, Zheng-Hua Tan

Detecting occurrences of keywords with keyword spotting (KWS) systems requires thresholding continuous detection scores. Selecting appropriate thresholds is a non-trivial task, typically relying on optimizing performance on a validation dataset. However, such greedy threshold selection often leads to suboptimal performance on unseen data, particularly in var

Hao Liu, Yiqing Dai, Haotian Tan, Yu Lei

Emotions guide human decisions, but whether large language models (LLMs) use emotion similarly remains unknown. We tested this using altruistic third-party punishment, where an observer incurs a personal cost to enforce fairness, a hallmark of human morality and often driven by negative emotion. In a large-scale comparison of 4,068 LLM agents with 1,159 adul

Francesco Colasanto, Pascal Steinke

This paper addresses the asymptotic development of order 2 by Gamma convergence of the Cahn-Hillard functional with Dirichlet boundary conditions, where the potential has subquadratic growth near the wells.

Shuang Liang, Zhihao Xu, Jialing Tao, Hui Xue

Despite extensive alignment efforts, Large Vision-Language Models (LVLMs) remain vulnerable to jailbreak attacks, posing serious safety risks. To address this, existing detection methods either learn attack-specific parameters, which hinders generalization to unseen attacks, or rely on heuristically sound principles, which limit accuracy and efficiency. To o

Shashank Gupta

This dissertation investigates how reinforcement learning (RL) methods can be designed to be safe, sample-efficient, and robust. Framed through the unifying perspective of contextual-bandit RL, the work addresses two major application domains - ranking and recommendation, and text-to-image diffusion models. The first part of the thesis develops theory and al

Mahta Khoobi, Marc Sebastian von der Stueck, Felix Barajas Ordonez, Anca-Maria Iancu

Structured reporting (SR) and artificial intelligence (AI) may transform how radiologists interact with imaging studies. This prospective study (July to December 2024) evaluated the impact of three reporting modes: free-text (FT), structured reporting (SR), and AI-assisted structured reporting (AI-SR), on image analysis behavior, diagnostic accuracy, efficie

S. O. Kara

We present the first quantitative demonstration of the statistical convergence of heavy-flavor parton distribution functions at small x across modern NNLO global fits. Using the latest NNPDF4.0, CT18, and MSHT20 sets, we find a maximal relative deviation of 16.8 percent at x = 1e-6 and Q = 100 GeV, corresponding to nearly a factor-of-two improvement over the

Yulong Zhang

We present OCR-Quality, a comprehensive human-annotated dataset designed for evaluating and developing OCR quality assessment methods. The dataset consists of 1,000 PDF pages converted to PNG images at 300 DPI, sampled from diverse real-world scenarios, including academic papers, textbooks, e-books, and multilingual documents. Each document has been processe

Kuan-Wei Ho, Yi-Hsin Chen, Martin Benjak, Jörn Ostermann

Recent advances in learned video codecs have demonstrated remarkable compression efficiency. Two fundamental design aspects are critical: the choice of inter-frame coding framework and the temporal information propagation strategy. Inter-frame coding frameworks include residual coding, conditional coding, conditional residual coding, and masked conditional r

Van Nam Dinh

Quadratic programming (QP) underpins real-time robotics by enabling efficient, constrained optimization in state estimation, motion planning, and control. In legged locomotion and manipulation, essential modules like inverse dynamics, Model Predictive Control (MPC), and Whole-Body Control (WBC) are inherently QP-based, demanding reliable solutions amid tight

Wei Wang, Xiao-Yong Wei, Qing Li

The widespread 'deeper is better' philosophy has driven the creation of architectures like ResNet and Transformer, which achieve high performance by stacking numerous layers. However, increasing model depth comes with challenges such as longer training times, higher inference latency, and impracticality on resource-constrained devices. To address these issue

Amir Mohammad Mirzaei

Predicting fracture load in laminated composites with stress raisers is challenging due to complex failure mechanisms such as delamination, fibre breakage, and matrix cracking, which are heavily influenced by fibre orientation, layup sequence, and notch geometry. This study aims to address this by developing a novel deep learning framework that leverages sol

Òscar Burés

In this paper we study the short-maturity asymptotics of up-and-in barrier options under a broad class of stochastic volatility models. Our approach uses Malliavin calculus techniques, typically used for linear stochastic partial differential equations, to analyse the law of the supremum of the log-price process. We derive a concentration inequality and expl

Lin Wang

Open world Machine Learning (OWML) aims to develop intelligent systems capable of recognizing known categories, rejecting unknown samples, and continually learning from novel information. Despite significant progress in open set recognition, novelty detection, and continual learning, the field still lacks a unified theoretical foundation that can quantify un

Zheyuan Lin, Siqi Cai, Haizhou Li

EEG-based person identification enables applications in security, personalized brain-computer interfaces (BCIs), and cognitive monitoring. However, existing techniques often rely on deep learning architectures at high computational cost, limiting their scope of applications. In this study, we propose a novel EEG person identification approach using spiking n

Hongcheng Liu, Pingjie Wang, Yuhao Wang, Siqu Ou

Multimodal large language models (MLLMs) have shown strong capabilities across a broad range of benchmarks. However, most existing evaluations focus on passive inference, where models perform step-by-step reasoning under complete information. This setup is misaligned with real-world use, where seeing is not enough. This raises a fundamental question: Can MLL

Shweta Bahl, Ajay Sharma

Using nationally representative data for India, this paper examines the incidence of education occupation mismatch and returns to education and EOM for internal migrants while considering the heterogeneity among them. In particular, this study considers heterogeneity arising because of the reason to migrate, demographic characteristics, spatial factors, migr

Hrithik Barman

We investigate the interplay between chirality and confinement in harmonically trapped active particles. The circular character of chiral motion combines with the radial symmetry of the potential to create distinctive non-equilibrium behavior. Chirality induces oscillatory cross-correlations between positional components that vanish in the absence of torque

Lee Qi Zun, Mohamad Zulhilmi Bin Abdul Halim, Goh Man Fye

Retrieval-Augmented Generation systems are essential for providing fact-based guidance from Malaysian Clinical Practice Guidelines. However, their effectiveness with image-based queries is limited, as general Vision-Language Model captions often lack clinical specificity and factual grounding. This study proposes and validates a framework to specialize the M

Xavier Buffat

Landau damping is a key mechanism to preserve the stability of particle beams under the influence of various collective forces that would otherwise spoil its quality through beam instabilities. We describe its root cause as well as ways to control it in order to design and operate particle accelerators.

Helena Khoury Moussa, Georges Challita, Houssem Badreddine, Guillaume Montay

Advances in technology have provided fresh generations of stiff polypropylene block copolymers for gravity sewerage applications. The aim of this study is to further enhance the stiffness of these materials through the incorporation of inorganic fillers. In this study, three talc filled PP and one glass fiber filled PP composites were characterized in order

Huining Yuan, Zelai Xu, Zheyue Tan, Xiangmin Yi

Developing Large Language Models (LLMs) to cooperate and compete effectively within multi-agent systems (MASs) is a critical step towards more advanced intelligence. While reinforcement learning (RL) has proven effective for enhancing reasoning in single-agent tasks, its extension to multi-turn, multi-agent scenarios remains underexplored due to the challeng

Po-Yu Tseng, Po-Chu Hsu, Shih-Wei Liao

FHE-SQL is a privacy-preserving database system that enables secure query processing on encrypted data using Fully Homomorphic Encryption (FHE), providing privacy guaranties where an untrusted server can execute encrypted queries without learning either the query contents or the underlying data. Unlike property-preserving encryption-based systems such as Cry