May 2025 arXiv papers — page 77

Julian Oelhaf, Georg Kordowich, Andreas Maier, Johann Jager

The widespread use of sensors in modern power grids has led to the accumulation of large amounts of voltage and current waveform data, especially during fault events. However, the lack of labeled datasets poses a significant challenge for fault classification and analysis. This paper explores the application of unsupervised clustering techniques for fault di

Ziyu Ge, Yuhao Wu, Daniel Wai Kit Chin, Roy Ka-Wei Lee

Large Language Models (LLMs) augmented with retrieval mechanisms have demonstrated significant potential in fact-checking tasks by integrating external knowledge. However, their reliability decreases when confronted with conflicting evidence from sources of varying credibility. This paper presents the first systematic evaluation of Retrieval-Augmented Genera

Benjamin Walker, Lingyi Yang, Nicola Muca Cirone, Cristopher Salvi

This work introduces Structured Linear Controlled Differential Equations (SLiCEs), a unifying framework for sequence models with structured, input-dependent state-transition matrices that retain the maximal expressivity of dense matrices whilst being cheaper to compute. The framework encompasses existing architectures, such as input-dependent block-diagonal

Leon Eshuijs, Archie Chaudhury, Alan McBeth, Ethan Nguyen

LLM-as-a-judge is widely used as a scalable substitute for human evaluation, yet current approaches rely on black-box access and struggle to detect subtle dishonesty, such as sycophancy and manipulation. We introduce Judge Using Safety-Steered Alternatives (JUSSA), a framework that leverages a model's internal representations to optimize an honesty-promoting

Roni Dey, Dipanwita Mondal, Sudipta Das, Varchaswi K. S. Kashyap

Detailed understanding and suppression of backgrounds are among the key challenges faced by Coherent Elastic Neutrino-Nucleus Scattering (CE\ensuremath{\nu}NS) experiments. The sensitivity of these experiments is largely determined by the background levels arising from various sources. Above-ground and shallow-overburden neutrino experiments typically employ

Chris Pearson, Thomas W. O. Varnish, Xinni Wu, David L. Clements

We present the image maps, data reduction, analysis and the first source counts from the Herschel SPIRE Dark Field. The SPIRE Dark Field is an area of sky near the North Ecliptic Pole observed many times during the calibration phase of the Herschel mission in order to characterise the stability of the SPIRE instrument and is subsequently one of the deepest i

Wang Chen, Hongzheng Shi, Jintao Ke

The rapid growth of ride-sharing services presents a promising solution to urban transportation challenges, such as congestion and carbon emissions. However, developing efficient operational strategies, such as pricing, matching, and fleet management, requires robust simulation tools that can replicate real-world dynamics at scale. Existing platforms often l

P. J. Costello, G. G. Plunk

Bounding energetic growth of gyrokinetic instabilities is a complementary approach to linear instability analyses involving normal eigenmodes. Previous work has focused on upper bounds which are valid linearly and nonlinearly. However, if an upper bound on linear instability growth is desired, these nonlinearly valid bounds may be a poor predictor of the gro

M. Emre Sahin, Edoardo Altamura, Oscar Wallis, Stephen P. Wood

We present Qiskit Machine Learning (ML), a high-level Python library that combines elements of quantum computing with traditional machine learning. The API abstracts Qiskit's primitives to facilitate interactions with classical simulators and quantum hardware. Qiskit ML started as a proof-of-concept code in 2019 and has since been developed to be a modular,

Dafni Skiadopoulou, Lukas Käll, Harald Barsnes, Veit Schwämmle

Background In proteomics, the most probable localizations of post-translational modifications are assessed by localization scores evaluating the likelihood of a given modification to occupy a site on a peptide sequence. When identifying highly modified peptides, localization scores for different modifications can return conflicting results, stacking modifica

Netzer Moriya

We present a new mathematical framework for incorporating partial coherence effects into wave optics simulations through a comprehensive surface-to-detector approach. Unlike traditional ensemble averaging methods, our dual-component framework models partial coherence through: (1) a surface-encoded transformation implemented via a linear integral operator wit

R Shivananda Rao, M Ramakrishna

We adapt a troubled-cell indicator from discontinuous Galerkin (DG) methods to finite volume methods (FVM) with MUSCL reconstruction and using a novel monotonicity parameter show there is a trade-off between convergence and quality of the solution. Employing two dimensional compressible Euler equations for flows with oblique shocks, this trade-off is studied

Andrea Malchiodi, Francesco Malizia

Given two closed, oriented Riemannian four-manifolds $(M,g_M)$ and $(Z,g_Z)$, which are not locally conformally flat and not both self-dual or both anti-self-dual, we prove that there exists a metric $g_Y$ on the connected sum $Y\cong M\#Z$ such that the Weyl energy of $g_Y$ is strictly smaller than the sum of Weyl energies of $g_M$ and $g_Z$.

Michael Feischl, Fabian Zehetgruber

We show that under some widely believed assumptions, there are no higher-order algorithms for basic tasks in computational mathematics such as: Computing integrals with neural network integrands, computing solutions of a Poisson equation with neural network source term, and computing the matrix-vector product with a neural network encoded matrix. We show tha

G. K. Goswami, Anirudh Pradhan, Syamala Krishnannair

We investigate a spatially flat FLRW cosmological model in the framework of modified gravity described by the function \( f(R, L_m) = \alpha R + L_m^\beta + \gamma \), where \( L_m \) is the matter Lagrangian density. The modified Friedmann equations yield the Hubble parameter as $ H(z) = H_0 \sqrt{(1 - \lambda) + \lambda (1 + z)^{3(1 + w)}},$ with the param

Vaisakh Plakkot

The possible Majorana nature of neutrinos leads to lepton-number-violating effects such as neutrinoless double beta decay. The standard study of this process involves mass-dependent matrix elements which, although easy to use, might be missing important effects, especially in the light neutrino regime where the ultrasoft contributions become important. A fre

Ghada Sokar, Pablo Samuel Castro

Scaling deep reinforcement learning in pixel-based environments presents a significant challenge, often resulting in diminished performance. While recent works have proposed algorithmic and architectural approaches to address this, the underlying cause of the performance drop remains unclear. In this paper, we identify the connection between the output of th

Kerol Djoumessi, Philipp Berens

Convolutional neural networks (CNNs) are widely used for high-stakes applications like medicine, often surpassing human performance. However, most explanation methods rely on post-hoc attribution, approximating the decision-making process of already trained black-box models. These methods are often sensitive, unreliable, and fail to reflect true model reason

Maureen de Seyssel, Jie Chi, Skyler Seto, Maartje ter Hoeve

We introduce a set of training-free ABX-style discrimination tasks to evaluate how multilingual language models represent language identity (form) and semantic content (meaning). Inspired from speech processing, these zero-shot tasks measure whether minimal differences in representation can be reliably detected. This offers a flexible and interpretable alter

Wei Huang, Yizhe Xiong, Xin Ye, Zhijie Deng

Large Language Models (LLMs) have achieved impressive performance across a range of natural language processing tasks. However, recent advances demonstrate that further gains particularly in complex reasoning tasks require more than merely scaling up model sizes or training data. One promising direction is to enable models to think during the reasoning proce

Zeyuan Ma, Yue-Jiao Gong, Hongshu Guo, Wenjie Qiu

Meta-Black-Box Optimization (MetaBBO) streamlines the automation of optimization algorithm design through meta-learning. It typically employs a bi-level structure: the meta-level policy undergoes meta-training to reduce the manual effort required in developing algorithms for low-level optimization tasks. The original MetaBox (2023) provided the first open-so

Amitai Yacobi, Nir Ben-Ari, Ronen Talmon, Uri Shaham

Learning shared representations is a primary area of multimodal representation learning. The current approaches to achieve a shared embedding space rely heavily on paired samples from each modality, which are significantly harder to obtain than unpaired ones. In this work, we demonstrate that shared representations can be learned almost exclusively from unpa

Chad Henshaw, Alice Heranval, Laura Cadonati

Gravitational wave signals from asymmetric binary black hole systems have been shown to exhibit additional chirps beyond the primary merger chirp in the post-merger region of the time-frequency domain. These secondary post-merger chirps correlate to the evolving geometry of the common horizon that forms as the binary merges and were previously studied throug

Yingzhuo Lun, Xinxin Hu, Qi Ren, Umair Saeed

Recent advances in moire engineering provide new pathways for manipulating lattice distortions and electronic properties in low-dimensional materials. Here, we demonstrate that twisted stacking can induce dipolar vortices in metallic SrRuO3 membranes, despite the presence of free charges that would normally screen depolarizing fields and dipole-dipole intera

Zijing Ou, Ruixiang Zhang, Yingzhen Li

Sampling from unnormalised discrete distributions is a fundamental problem across various domains. While Markov chain Monte Carlo offers a principled approach, it often suffers from slow mixing and poor convergence. In this paper, we propose Discrete Neural Flow Samplers (DNFS), a trainable and efficient framework for discrete sampling. DNFS learns the rate

Fengyi Wang, Xiangyu Fu, Nitish Thakor, Gordon Cheng

Proprioception, a key sensory modality in haptic perception, plays a vital role in perceiving the 3D structure of objects by providing feedback on the position and movement of body parts. The restoration of proprioceptive sensation is crucial for enabling in-hand manipulation and natural control in the prosthetic hand. Despite its importance, proprioceptive

Nasim Alikhani, Abbas Mohammadi

This paper introduces an optimum solution for a utility function that increases spectral efficiency in wireless Virtual Reality (VR) systems. This system uses Multi-user Multiple Input Multiple Output Orthogonal Frequency Division Multiplexing (MU-MIMO OFDM) with hybrid beamforming in indoor Intelligent Reflecting Surface (IRS) based Downlink (DL) scenario.

Nilanjan Sinhababu, Andrew Parry, Debasis Ganguly, Pabitra Mitra

While standard IR models are primarily designed to optimize relevance, real-world search often needs to balance additional objectives such as diversity and fairness. These objectives depend on inter-document interactions and are commonly addressed using post-hoc heuristics or supervised learning methods, which require task-specific training for each ranking

Xueyang Zhou, Weidong Wang, Lin Lu, Jiawen Shi

Large Language Model (LLM)-based agents are increasingly deployed in real-world applications such as "digital assistants, autonomous customer service, and decision-support systems", where their ability to "interact in multi-turn, tool-augmented environments" makes them indispensable. However, ensuring the safety of these agents remains a significant challeng

Ahmet Onur Akman, Anastasia Psarou, Michał Hoffmann, Łukasz Gorczyca

Connected Autonomous Vehicles (CAVs) promise to reduce congestion in future urban networks, potentially by optimizing their routing decisions. Unlike for human drivers, these decisions can be made with collective, data-driven policies, developed using machine learning algorithms. Reinforcement learning (RL) can facilitate the development of such collective r

Maria Petrova, Maria Ponomareva, Alexandra Ivoylova

The paper is devoted to the creation of the semantic sketches for English verbs. The pilot corpus consists of the English-Russian sketch pairs and is aimed to show what kind of contrastive studies the sketches help to conduct. Special attention is paid to the cross-language differences between the sketches with similar semantics. Moreover, we discuss the pro

Ozsel Kilinc, Cem Tarhan

Accurate, fast, and reliable 3D perception is essential for autonomous driving. Recently, bird's-eye view (BEV)-based perception approaches have emerged as superior alternatives to perspective-based solutions, offering enhanced spatial understanding and more natural outputs for planning. Existing BEV-based 3D object detection methods, typically using an angl

Adam Bílek, Jan Hlisnikovský, Tomáš Bezděk, Ryszard Kukulski

Tasks involving black boxes appear frequently in quantum computer science. An example that has been deeply studied is quantum channel discrimination. In this work, we study the discrimination between two quantum unitary channels in the multiple-shot scenario. We challenge the theoretical results concerning the probability of correct discrimination with the r

Stefan Schoepf, Michael Curtis Mozer, Nicole Elyse Mitchell, Alexandra Brintrup

Machine unlearning is studied for a multitude of tasks, but specialization of unlearning methods to particular tasks has made their systematic comparison challenging. To address this issue, we propose a conceptual space to characterize diverse corrupted data unlearning tasks in vision classifiers. This space is described by two dimensions, the discovery rate

J. Vaz, E. Capelas de Oliveira

In this paper we discuss some issues that arise in the process of writing a fractional differential equation (FDE) by replacing an integer order derivative by a fractional order derivative in a given differential equation. To address these issues, we propose a dimensional regularization of the Caputo fractional derivative, ensuring consistency in physical di

Chiara Esposito, Andrea Rivezzi, Jonas Schnitzer, Thomas Weber

In this paper we extend Cartier's deformation theorem of braided monoidal categories admitting an infinitesimal braiding to the non-symmetric case. The algebraic counterpart of these categories is the notion of a pre-Cartier quasi-bialgebra, which extends the well-known notion of quasitriangular quasi-bialgebra given by Drinfeld. Our result implies that one

Maciej Nytko, Mateusz Kędziora, Maciej Zaremba, Marek Ekielski

Perovskite microwires are promising candidates for integrated photonic systems due to their strong nonlinear optical response and inherent waveguiding capabilities. In this study, we focus on the directional emission properties of exciton-polariton condensates formed within perovskite microwires, with emphasis on emission collected from the microwire end. We

Jiawei Zhou, Linye Lyu, Zhuotao Tian, Cheng Zhuo

Safety-critical scenarios are rare yet pivotal for evaluating and enhancing the robustness of autonomous driving systems. While existing methods generate safety-critical driving trajectories, simulations, or single-view videos, they fall short of meeting the demands of advanced end-to-end autonomous systems (E2E AD), which require real-world, multi-view vide

Gerhard Schindl

We apply recent knowledge and techniques of the new generalized upper and lower Legendre conjugates to the theory of weight functions in the sense of Braun-Meise-Taylor and study in detail the effects on the corresponding associated weight matrices. An immediate and concrete application of the main statements is also provided. More precisely, we generalize a

Fengyi Wang, J. Rogelio Guadarrama Olvera, Nitish Thako, Gordon Cheng

The nociceptive withdrawal reflex (NWR) is a mechanism to mediate interactions and protect the body from damage in a potentially dangerous environment. To better convey warning signals to users of prosthetic arms or autonomous robots and protect them by triggering a proper NWR, it is useful to use a biological representation of temperature information for fa

Benoît Jeanson, Simon H. Tindemans

This paper deals with the secure Optimal Transmission Switching (OTS) problem in situations where the TSO is forced to accept the risk that some contingencies may result in the de-energization of parts of the grid to avoid the violation of operational limits. This operational policy, which mainly applies to subtransmission systems, is first discussed. Then,

Rosa Flaquer-Galmés, Daniel Campos, Javier Cristín

We propose a physical framework for ant navigation of chemical trails. For this, we use controlled experiments in which individuals follow narrow pheromone trails, for which ants display oscillatory motion, as previously reported in the literature. We model this behavior by treating chemotaxis as an effective magnetic interaction between the ant velocity and

Dekai Zhu, Yan Di, Stefan Gavranovic, Slobodan Ilic

Denoising diffusion probabilistic models have achieved significant success in point cloud generation, enabling numerous downstream applications, such as generative data augmentation and 3D model editing. However, little attention has been given to generating point clouds with point-wise segmentation labels, as well as to developing evaluation metrics for thi

Fabio Durastante, Mariarosa Mazza

Implicit Runge--Kutta (IRK) methods are highly effective for solving stiff ordinary differential equations (ODEs) but can be computationally expensive for large-scale problems due to the need of solving coupled algebraic equations at each step. This study improves IRK efficiency by leveraging parallelism to decouple stage computations and reduce communicatio

Josu Etxezarreta Martinez, Paul Schnabl, Javier Oliva del Moral, Reza Dastbasteh

Tailoring quantum error correction codes (QECC) to biased noise has demonstrated significant benefits. However, most of the prior research on this topic has focused on code capacity noise models. Furthermore, a no-go theorem prevents the construction of CNOT gates for two-level qubits in a bias preserving manner which may, in principle, imply that noise bias

Akira Tanimoto

Causal inference requires evaluating models on balanced distributions between treatment and control groups, while training data often exhibits imbalance due to historical decision-making policies. Most conventional statistical methods address this distribution shift through inverse probability weighting (IPW), which requires estimating propensity scores as a

Erhu Feng, Wenbo Zhou, Zibin Liu, Le Chen

AI agents, empowered by Large Language Models (LLMs) and communication protocols such as MCP and A2A, have rapidly evolved from simple chatbots to autonomous entities capable of executing complex, multi-step tasks, demonstrating great potential. However, the LLMs' inherent uncertainty and heavy computational resource requirements pose four significant challe

Carles López-Mateu, Maria Gómez-Mahiques, F. Javier Gil-Terrón, Víctor Montosa-i-Micó

Glioblastoma (GBM) exhibits two principal growth phenotypes: infiltrative, characterized by diffuse invasion with minimal mass effect, and proliferative, characterized by pronounced tissue compression. Their quantitative delineation and prognostic implications remain uncertain. We introduce an MRI-derived biomarker, the dynamic infiltration rate (DIR), defin

M. Ali Allousch, André Sandmann

CC-OTDR signal envelope shaping is introduced to reduce the impact of non-linear signal interactions on a neighboring wavelength data channel when co-propagating the probing signal with the data signal. Joint co-directional acoustic sensing and 200 Gbps transmission are demonstrated over a 50 km link.

Ben Rahman

Despite Proximal Policy Optimization (PPO) dominating policy gradient methods -- from robotic control to game AI -- its static trust region forces a brittle trade-off: aggressive clipping stifles early exploration, while late-stage updates destabilize convergence. PPO-BR establishes a new paradigm in adaptive RL by fusing exploration and convergence signals

Frans Perkkola, Ilmo Salmeperä, Arianne Meijer-van de Griend, C. -C. Joseph Wang

The execution of quantum algorithms on modern hardware is often constrained by noise and qubit decoherence, limiting the circuit depth and the number of gates that can be executed. Circuit optimization techniques help mitigate these limitations, enhancing algorithm feasibility. In this work, we implement, optimize, and execute a variational quantum regressio

Yi Su, Jiayi Zhang, Shu Yang, Xinhai Wang

Rapid integration of large language models (LLMs) into societal applications has intensified concerns about their alignment with universal ethical principles, as their internal value representations remain opaque despite behavioral alignment advancements. Current approaches struggle to systematically interpret how values are encoded in neural architectures,

Misha Khalid, Ankur Chatterjee, Ewa Przezdziecka, Abinash Adhikari

This study elucidates the intriguing phenomenon of inverse thermal anisotropy in cadmium magnesium oxide (CdMgO) thin films, characterized by cross-plane thermal conductivity being greater than in-plane thermal conductivity, essential for optimizing thermal management in next-generation optoelectronic devices. Herein, we utilized Photothermal Radiometry and

Rafael Corsi Ferrao, Fabio Roberto de Miranda, Diego Pavan Soler

This full paper in innovative practice provides an automated tool to summarize individual code contributions in project-based courses with external clients. Real industry projects offer valuable learning opportunities by immersing students in authentic problems defined by external clients. However, the open-ended and highly variable scope of these projects m

Harald Havir, Andrea Cicovic, Pierre Glidic, Subhomoy Haldar

Dissipative sensors typically use linear resonators with impedance matching to achieve maximal signal and fast operation. The impedance matching, however, sets an upper limit to the bandwidth of the readout. In this paper, we present a nonlinear resonator performing the readout of a double quantum dot charge state via a charge-sensing quantum dot. We show th

Dingling Yao, Shimeng Huang, Riccardo Cadei, Kun Zhang

Causal reasoning and discovery, two fundamental tasks of causal analysis, often face challenges in applications due to the complexity, noisiness, and high-dimensionality of real-world data. Despite recent progress in identifying latent causal structures using causal representation learning (CRL), what makes learned representations useful for causal downstrea

Ariane Hanebeck, Özge Şahin, Petra Havlíčková, Claudia Czado

Simplified vine copulas are flexible tools over standard multivariate distributions for modeling and understanding different dependence properties in high-dimensional data. Their conditional distributions are of utmost importance, from statistical learning to graphical models. However, the conditional densities of vine copulas and, thus, vine distributions c

Runze Li, Siyu Wu, Jun Wang, Wei Zhang

Knowledge Tracing (KT) aims to model a student's learning state over time and predict their future performance. However, traditional KT methods often face challenges in explainability, scalability, and effective modeling of complex knowledge dependencies. While Large Language Models (LLMs) present new avenues for KT, their direct application often struggles

Maria Ponomareva, Maria Petrova, Julia Detkova, Oleg Serikov

The paper deals with elaborating different approaches to the machine processing of semantic sketches. It presents the pilot open corpus of semantic sketches. Different aspects of creating the sketches are discussed, as well as the tasks that the sketches can help to solve. Special attention is paid to the creation of the machine processing tools for the corp

André Silva, Gustav Thorén, Martin Monperrus

Automatic program repair seeks to generate correct code from buggy programs, with most approaches searching the correct program in a discrete, symbolic space of source code tokens. This symbolic search is fundamentally limited by its inability to directly reason about program behavior. We introduce Gradient-Based Program Repair (GBPR), a new approach that re

Xueyang Li, Jiahao Li, Yu Song, Yunzhong Lou

The advent of Computer-Aided Design (CAD) generative modeling will significantly transform the design of industrial products. The recent research endeavor has extended into the realm of Large Language Models (LLMs). In contrast to fine-tuning methods, training-free approaches typically utilize the advanced closed-source LLMs, thereby offering enhanced flexib

Jaewon Cheon, Pilsung Kang

The growing size of large language models has created significant computational inefficiencies. To address this challenge, sparse activation methods selectively deactivates non-essential parameters during inference, reducing computational costs in FFNN layers. While existing methods focus on non-linear gating mechanisms, we hypothesize that the sparsity of t

Elisa Ballin, Francesco Brasili, Tommaso Papetti, Jacopo Vialetto

We provide experimental and numerical evidence of a new class of soft nanocolloids: star-like microgels with thermoresponsive character. This is achieved by using the standard precipitation polymerization synthesis of poly(N-isopropylacrylamide) (PNIPAM) microgels and replacing the usually employed crosslinking agent, N,N'-methylenebis(acrylamide) (BIS), wit

Rémi Pallen, Ilkka Törmä

We define sets of coulourings of the infinite discrete plane using monadic second order (MSO) formulas. We determine the complexity of deciding whether such a formula defines a subshift, parametrized on the quantifier alternation complexity of the formula. We also study the complexities of languages of MSO-definable sets, giving either an exact classificatio

Harshita Devda, András Deák, Leandro Salemi, Levente Rózsa

Understanding the mechanism driving magnetization switching in spin-orbit-torque-assisted devices remains a subject of debate. While originally attributed to the spin Hall effect and spin Rashba-Edelstein effect, recent discoveries related to orbital moments induced by the orbital Hall effect and the orbital Rashba-Edelstein effect have added complexity to t

Zekai Zhao, Qi Liu, Kun Zhou, Zihan Liu

Despite the remarkable reasoning performance, eliciting the long chain-of-thought (CoT) ability in large language models (LLMs) typically requires costly reinforcement learning or supervised fine-tuning on high-quality distilled data. We investigate the internal mechanisms behind this capability and show that a small set of high-impact activations in the las

Sota Yoshihara, Ryosuke Yamamoto, Hiroyuki Kusumoto, Masanari Shimura

This paper proposes a novel theoretical framework for guaranteeing and evaluating the resilience of long short-term memory (LSTM) networks in control systems. We introduce "recovery time" as a new metric of resilience in order to quantify the time required for an LSTM to return to its normal state after anomalous inputs. By mathematically refining incrementa

Dong-Hee Kim, Hyunjee Song, Donghyun Kim

Despite the advances in Referring Expression Segmentation (RES) benchmarks, their evaluation protocols remain constrained, primarily focusing on either single targets with short queries (containing minimal attributes) or multiple targets from distinctly different queries on a single domain. This limitation significantly hinders the assessment of more complex

Gian Maria Dall'Ara, Samuele Mongodi, John N. Treuer

We introduce the Grassmannian $q$-core of a distribution of subspaces of the tangent bundle of a smooth manifold. This is a generalization of the concept of the core previously introduced by the first two authors. In the case where the distribution is the Levi null distribution of a smooth bounded pseudoconvex domain $\Omega\subseteq \mathbb{C}^n$, we prove

Ziteng Yang, Jingzehua Xu, Yanshu Li, Zepeng Li

Zero-shot anomaly detection (ZSAD) aims to detect anomalies without any target domain training samples, relying solely on external auxiliary data. Existing CLIP-based methods attempt to activate the model's ZSAD potential via handcrafted or static learnable prompts. The former incur high engineering costs and limited semantic coverage, whereas the latter app

Yan Yu, Yilun Liu, Minggui He, Shimin Tao

Pairwise evaluation of large language models (LLMs) has become the dominant paradigm for benchmarking open-ended tasks, yet non-transitive preferences, where evaluators prefer A over B, B over C, but C over A, fundamentally undermine ranking reliability. We show that this critical issue stems largely from low-quality data that contains inherently ambiguous p

Yunyao Lu, Yihang Wu, Reem Kateb, Ahmad Chaddad

Traditional supervised medical image segmentation models require large amounts of labeled data for training; however, obtaining such large-scale labeled datasets in the real world is extremely challenging. Recent semi-supervised segmentation models also suffer from noisy pseudo-label issue and limited supervision in feature space. To solve these challenges,

Bocquet Jean-Louis

Calculating the vibrational entropy of an N-atom assembly in the harmonic approximation requires the diagonalisation of a large matrix. This operation becomes rapidly time consuming when increasing the dimensions of the simulation cell. In the studies of point defects, a widely used shortcut consists in calculating the eigen modes of the atoms contained in a

Elia Moretti, Michel Loreau, Michael Benzaquen

The intensification of European agriculture, characterized by increasing farm sizes, landscape simplification and reliance on synthetic pesticides, remains a key driver of biodiversity decline. While many studies have investigated this phenomenon, they often focus on isolated elements, resulting in a lack of holistic understanding and leaving policymakers an

H. Abele, G. Angloher, B. Arnold, M. Atzori Corona

Percent-level calibration of cryogenic macro-calorimeters with energy thresholds below 100~eV are crucial for light Dark Matter (DM) searches and reactor neutrino studies based on coherent elastic neutrino-nucleus scattering (CEvNS). This paper presents a novel calibration source based on X-ray fluorescence (XRF) of light elements. It uses a $^{55}$Fe source

Shuang Zeng, Xinyuan Chang, Mengwei Xie, Xinran Liu

Vision-Language-Action (VLA) models offer significant potential for end-to-end driving, yet their reasoning is often constrained by textual Chains-of-Thought (CoT). This symbolic compression of visual information creates a modality gap between perception and planning by blurring spatio-temporal relations and discarding fine-grained cues. We introduce FSDrive

Nisha Lakshmana Raichur, Lucas Heublein, Christopher Mutschler, Felix Ott

Indoor positioning based on 5G data has achieved high accuracy through the adoption of recent machine learning (ML) techniques. However, the performance of learning-based methods degrades significantly when environmental conditions change, thereby hindering their applicability to new scenarios. Acquiring new training data for each environmental change and fi

Dan Yuan, Yi Feng, Ziyun Tang

Intraventricular hemorrhage (IVH) is a severe neurological complication among premature infants, necessitating early and accurate detection from brain ultrasound (US) images to improve clinical outcomes. While recent deep learning methods offer promise for computer-aided diagnosis, challenges remain in capturing both local spatial details and global contextu

Fanjin Meng, Jingtao Ding, Jiahui Gong, Chen Yang

Predicting user behavior is essential for intelligent assistant services, yet deep learning models often struggle to capture long-tailed behaviors. Large language models (LLMs), with their pretraining on vast corpora containing rich behavioral knowledge, offer promise. However, existing fine-tuning approaches tend to overfit to frequent ``anchor'' behaviors,

P. S. Bhupal Dev

This is a brief review of the collider phenomenology of neutrino physics. Current and future colliders provide an ideal testing ground for (sub)TeV-scale neutrino mass models, as they can directly probe the messenger particles, which could be either new fermions, scalars, or gauge bosons, associated with neutrino mass generation. Moreover, the recent observa

D. Langemann, A. S. Mikhaylov, V. S. Mikhaylov

We consider the problem of reconstruction of Cauchy data for the wave equation in $\mathbb{R}^1$ by the measurements of its solution on the boundary of the finite interval. This is a one-dimensional model for the multidimensional problem of photoacoustics, which was studied in \cite{BLMM}. We adapt and simplify the method for one-dimensional situation and pr

Ming Hu, Zhengdi Yu, Feilong Tang, Kaiwen Chen

Accurate 3D reconstruction of hands and instruments is critical for vision-based analysis of ophthalmic microsurgery, yet progress has been hampered by the lack of realistic, large-scale datasets and reliable annotation tools. In this work, we introduce OphNet-3D, the first extensive RGB-D dynamic 3D reconstruction dataset for ophthalmic surgery, comprising

M. Litniewski, W. T. Gozdz nd A. Ciach

Lateral microsegregation in a monolayer of a binary mixture of particles or macromolecules is studied by MD simulations in a generic model with the interacting potentials inspired by effective interactions in biological or soft-matter systems. In the model, the energy is minimized when like particles form small clusters, and the cross-interction is of opposi

A. S. Mikhaylov, V. S. Mikhaylov

We consider the inverse dynamic problem for the wave equation with a potential on an interval $(0,2\pi)$ with periodic boundary conditions. We use a boundary triplet to set up the initial-boundary value problem. As an inverse data we use a response operator (dynamic Dirichlet-to-Neumann map). Using the auxiliary problem on the whole line, we derive equations

Jiawei Du, Jinlong Wu, Yuzheng Chen, Yucheng Hu

Most LLM-based agent frameworks adopt a top-down philosophy: humans decompose tasks, define workflows, and assign agents to execute each step. While effective on benchmark-style tasks, such systems rely on designer updates and overlook agents' potential to learn from experience. Recently, Silver and Sutton(2025) envision a shift into a new era, where agents

Michael A. Henning, Dimbinaina Ralaivaosaona

A set $S$ of vertices in a graph $G$ is a paired dominating set if every vertex of $G$ is adjacent to a vertex in $S$ and the subgraph induced by $S$ contains a perfect matching (not necessarily as an induced subgraph). The paired domination number, $\gamma_{\mathrm{pr}}(G)$, of $G$ is the minimum cardinality of a paired dominating set of $G$. We present a l

Hanyu Li, Haoyu Liu, Tingyu Zhu, Tianyu Guo

Large Language Models (LLMs) show promise as data analysis agents, but existing benchmarks overlook the iterative nature of the field, where experts' decisions evolve with deeper insights of the dataset. To address this, we introduce IDA-Bench, a novel benchmark evaluating LLM agents in multi-round interactive scenarios. Derived from complex Kaggle notebooks

Yilun Liu, Chunguang Zhao, Xinhua Yang, Hongyong Zeng

Despite doubts on data quality, instruction synthesis has been widely applied into instruction tuning (IT) of LLMs as an economic and rapid alternative. Recent endeavors focus on improving data quality for synthesized instruction pairs in English and have facilitated IT of English-centric LLMs. However, data quality issues in multilingual synthesized instruc

Wenyi Wu, Zixuan Song, Kun Zhou, Yifei Shao

Language models (LMs) and their extension, vision-language models (VLMs), have achieved remarkable performance across various tasks. However, they still struggle with complex reasoning tasks that require multimodal or multilingual real-world knowledge. To support such capabilities, an external memory system that can efficiently provide relevant multimodal in

Raúl Hidalgo-Sacoto, Thomas Busch, D. Blume

Non-relativistic anyons in 1D possess generalized exchange statistics in which the exchange of two identical anyons generates a non-local phase that is governed by the spatial ordering of the particles and the statistical parameter $\alpha$. Working in the continuum, we demonstrate the existence of two distinct types of 1D anyons, namely bosonic anyons and f

A. S. Mikhaylov, V. S. Mikhaylov

We consider the inverse dynamic problem for the wave equation with a potential on a real line. The forward initial-boundary value problem is set up with a help of boundary triplets. As an inverse data we use an analog of a response operator (dynamic Dirichlet-to-Neumann map). We derive equations of inverse problem and also point out the relationship between

Fanqi Wan, Weizhou Shen, Shengyi Liao, Yingcheng Shi

Recent large reasoning models (LRMs) have demonstrated strong reasoning capabilities through reinforcement learning (RL). These improvements have primarily been observed within the short-context reasoning tasks. In contrast, extending LRMs to effectively process and reason on long-context inputs via RL remains a critical unsolved challenge. To bridge this ga

Shuxian Ma, Zihao Dong, Runmin Cong, Sam Kwong

Deep learning-based multi-view coarse-grained 3D shape classification has achieved remarkable success over the past decade, leveraging the powerful feature learning capabilities of CNN-based and ViT-based backbones. However, as a challenging research area critical for detailed shape understanding, fine-grained 3D classification remains understudied due to th

Yichun Yu, Yuqing Lan, Zhihuan Xing, Xiaoyi Yang

High-resolution remote sensing (HRRS) image segmentation is challenging due to complex spatial layouts and diverse object appearances. While CNNs excel at capturing local features, they struggle with long-range dependencies, whereas Transformers can model global context but often neglect local details and are computationally expensive.We propose a novel appr

Jędrzej Kozal, Jan Wasilewski, Alif Ashrafee, Bartosz Krawczyk

Memorization impacts the performance of deep learning algorithms. Prior works have studied memorization primarily in the context of generalization and privacy. This work studies the memorization effect on incremental learning scenarios. Forgetting prevention and memorization seem similar. However, one should discuss their differences. We designed extensive e

Yang Xiao, Jiashuo Wang, Qiancheng Xu, Changhe Song

As Large Language Models (LLMs) increasingly participate in human-AI interactions, evaluating their Theory of Mind (ToM) capabilities - particularly their ability to track dynamic mental states - becomes crucial. While existing benchmarks assess basic ToM abilities, they predominantly focus on static snapshots of mental states, overlooking the temporal evolu

Tianheng Ling, Chao Qian, Lukas Johannes Haßler, Gregor Schiele

Transformer-based models have shown strong performance across diverse time-series tasks, but their deployment on resource-constrained devices remains challenging due to high memory and computational demand. While prior work targeting Microcontroller Units (MCUs) has explored hardware-specific optimizations, such approaches are often task-specific and limited

Marcel Binz, Akshay K. Jagadish, Milena Rmus, Eric Schulz

We introduce automated scientific minimization of regret (ASMR) -- a framework for automated computational cognitive science. Building on the principles of scientific regret minimization, ASMR leverages Centaur -- a recently proposed foundation model of human cognition -- to identify gaps in an interpretable cognitive model. These gaps are then addressed thr

Chenyang Li, Jinsong Chen, John E. Hopcroft, Kun He

Neighborhood-aware tokenized graph Transformers have recently shown great potential for node classification tasks. Despite their effectiveness, our in-depth analysis of neighborhood tokens reveals two critical limitations in the existing paradigm. First, current neighborhood token generation methods fail to adequately capture attribute correlations within a

F. O. Schumann, M. Paleschke, J. Henk, W. Widdra

Imaging of magnetic domains with a photoelectron emission microscope operated with photon energies in the threshold regime often suffers from low contrast. In this work we show by symmetry considerations, photoemission calculations, and imaging experiments, how the contrast can be improved significantly. The key to both domain selectivity and sizable intensi