May 2025 arXiv papers — page 11

Stefano Marseglia

Honda and Tate showed that the isogeny classes of abelian varieties of dimension $g$ over a finite field $\mathbb{F}_q$ are classified in terms of $q$-Weil polynomials of degree $2g$, that is, monic integer polynomials whose set of complex roots consists of $g$ conjugate pairs of absolute value $\sqrt{q}$. There are descriptions of the space of such polynomi

Shota Horiguchi, Atsushi Ando, Marc Delcroix, Naohiro Tawara

End-to-end speaker diarization enables accurate overlap-aware diarization by jointly estimating multiple speakers' speech activities in parallel. This approach is data-hungry, requiring a large amount of labeled conversational data, which cannot be fully obtained from real datasets alone. To address this issue, large-scale simulated data is often used for pr

Wei Zhong, Manasa Bharadwaj, Yixiao Wang, Yipeng Ji

Speculative decoding (SD) is a widely adopted approach for accelerating inference in large language models (LLMs), particularly when the draft and target models are well aligned. However, state-of-the-art SD methods typically rely on tightly coupled, self-attention-based Transformer decoders, often augmented with auxiliary pooling or fusion layers. This coup

Felipe Avila, Alexander Bonilla Rivera, Rafael C. Nunes, R. F. L. Holanda

In this work, we perform a statistical inference of the classical background law governing the evolution of the temperature of the cosmic microwave background radiation (CMB), given by $T_{\rm CMB}(z) = T_0(1 + z)$. To this end, we employ Gaussian Process (GP) regression techniques to reconstruct the temperature evolution based on two observational datasets:

A. Friesen, Yu. Kalinovsky, A. Khmelev

In this work we study the meson properties in the framework of an effective quark model. We start from the Bethe-Salpeter equation choosing the interaction kernel in nonlocal form with the Gaussian meson vertex function, characterized by a meson size parameter $\Lambda_H$. We demonstrate the model's predictive power by applying it to both light and heavy sys

Xin He, Xumeng Han, Longhui Wei, Lingxi Xie

Multimodal large language models (MLLMs) require a nuanced interpretation of complex image information, typically leveraging a vision encoder to perceive various visual scenarios. However, relying solely on a single vision encoder to handle diverse task domains proves difficult and inevitably leads to conflicts. Recent work enhances data perception by direct

Arda Özdoğru, Sergey Karpov, Asen Christov, Stanislav Vítek

Scientific CMOS (sCMOS) image sensors are a modern alternative to typical CCD detectors and are rapidly gaining popularity in observational astronomy due to their large sizes, low read-out noise, high frame rates, and cheap manufacturing. However, numerous challenges remain in using them due to fundamental differences between CCD and CMOS architectures, espe

Orfeas Menis Mastromichalakis, Jason Liartis, Kristina Rose, Antoine Isaac

Cultural Heritage (CH) data hold invaluable knowledge, reflecting the history, traditions, and identities of societies, and shaping our understanding of the past and present. However, many CH collections contain outdated or offensive descriptions that reflect historical biases. CH Institutions (CHIs) face significant challenges in curating these data due to

Mario Alviano, Wolfgang Faber, Luis Angel Rodriguez Reiners

We present ASP Chef Mustache, an extension of ASP Chef that enhances template-based rendering of ASP solutions using a logic-less templating system inspired by Mustache. Our approach integrates data visualization frameworks such as Tabulator, Chart.js, and vis.js, enabling interactive representations of ASP interpretations as tables, charts, and graphs. Must

Chaohui Xu, Qi Cui, Chip-Hong Chang

The pervasion of large-scale Deep Neural Networks (DNNs) and their enormous training costs make their intellectual property (IP) protection of paramount importance. Recently introduced passport-based methods attempt to steer DNN watermarking towards strengthening ownership verification against ambiguity attacks by modulating the affine parameters of normaliz

Narmeen Oozeer, Luke Marks, Shreyans Jain, Fazl Barez

Controlling multiple behavioral attributes in large language models (LLMs) at inference time is a challenging problem due to interference between attributes and the limitations of linear steering methods, which assume additive behavior in activation space and require per-attribute tuning. We introduce K-Steering, a unified and flexible approach that trains a

Florian Frantzen, Michael T. Schaub

In this paper, we propose HLSAD, a novel method for detecting anomalies in time-evolving simplicial complexes. While traditional graph anomaly detection techniques have been extensively studied, they often fail to capture changes in higher-order interactions that are crucial for identifying complex structural anomalies. These higher-order interactions can ar

Mahesh Godavarti

We introduce a novel framework consisting of a class of algebraic structures that generalize one-dimensional monoidal systems into higher dimensions by defining per-axis composition operators subject to non-commutativity and a global interchange law. These structures, defined recursively from a base case of vector-matrix pairs, model directional composition

Ali Khoramfar, Ali Ramezani, Mohammad Mahdi Mohajeri, Mohammad Javad Dousti

While Large Language Models (LLMs) achieve near-human performance on standard benchmarks, their capabilities often fail to generalize to complex, real-world problems. To bridge this gap, we introduce DeepQuestion, a scalable, automated framework that systematically elevates the cognitive complexity of existing datasets. Grounded in Bloom's taxonomy, DeepQues

Sagar Ghosh, Kushal Bose, Swagatam Das

Despite their central role in the success of foundational models and large-scale language modeling, the theoretical foundations governing the operation of Transformers remain only partially understood. Contemporary research has largely focused on their representational capacity for language comprehension and their prowess in in-context learning, frequently u

Jesús A. Álvarez López, Alejandro O. Majadas-Moure, David Mosquera-Lois

We introduce a theory of integration with respect to the fixed point index, offering a substantial improvement over previous approaches based on the Lefschetz number. This framework eliminates several restrictive assumptions -- such as the need for definability, openness, or f-invariance of subspaces -- thereby allowing broader applicability. We also present

Alberto Bordino, Thomas B. Berrett

We propose the density ratio permutation test, a hypothesis test that assesses whether the ratio between two densities is proportional to a known function based on independent samples from each distribution. The test uses an efficient Markov Chain Monte Carlo scheme to draw weighted permutations of the pooled data, yielding exchangeable samples and finite sa

Egil Diau

A central challenge in economics and artificial intelligence is explaining how financial behaviors-such as credit, insurance, and trade-emerge without formal institutions. We argue that these functions are not products of institutional design, but structured extensions of a single behavioral substrate: reciprocity. Far from being a derived strategy, reciproc

Duaa Kareem Qasim, Sabah Abdulazeez Jebur, Lafta Raheem Ali, Abdul Jalil M. Khalaf

Retinal diseases such as Diabetic Retinopathy (DR) and Macular Hole (MH) significantly impact vision and affect millions worldwide. Early detection is crucial, as DR, a complication of diabetes, damages retinal blood vessels, potentially leading to blindness, while MH disrupts central vision, affecting tasks like reading and facial recognition. This paper em

Simone Cammarasana, Giuseppe Patanè

The paper introduces the weighted convolution, a novel approach to the convolution for signals defined on regular grids (e.g., 2D images) through the application of an optimal density function to scale the contribution of neighbouring pixels based on their distance from the central pixel. This choice differs from the traditional uniform convolution, which tr

Tomasz Kobos

Let $n \geq 2$ be an integer such that an equiangular set of vectors $w_1, \ldots, w_d$ of the maximal possible cardinality (in relation to the the general Gerzon upper bound) exists in $\mathbb{K}^n$, where $\mathbb{K}=\mathbb{R}$ or $\mathbb{K}=\mathbb{C}$ (i.e. $d=\frac{n(n+1)}{2}$ in the real and $d=n^2$ in the complex case). We provide a complete charac

Marcell Fekete, Nathaniel R. Robinson, Ernests Lavrinovics, E. Djeride Jean-Baptiste

Cross-lingual transfer from related high-resource languages is a well-established strategy to enhance low-resource language technologies. Prior work has shown that adapters show promise for, e.g., improving low-resource machine translation (MT). In this work, we investigate an adapter souping method combined with cross-attention fine-tuning of a pre-trained

Callum Berry

We give details of a new isolated symplectic singularity found in an affine chart in a crepant partial resolution of $\mathbb{C}^4/G_5$, which is 4-dimensional, isolated, and locally simply-connected. We distinguish the new singularity among all known such by the fact that the projective tangent cone at the singularity is non-reduced. We also find all 12 of

Andrea Pedrotti, Michele Papucci, Cristiano Ciaccio, Alessio Miaschi

Recent advancements in Generative AI and Large Language Models (LLMs) have enabled the creation of highly realistic synthetic content, raising concerns about the potential for malicious use, such as misinformation and manipulation. Moreover, detecting Machine-Generated Text (MGT) remains challenging due to the lack of robust benchmarks that assess generaliza

Andrés E. Piatti

The inner Milky Way disk globular cluster NGC~6362 appears to exhibit tidal tails composed of stars that have proper motions and positions in the color-magnitude diagram similar to those of cluster stars. Because recent results seem also to show that these stars are distributed across the regions least affected by interstellar absorption and reproduce the ob

Yang-Tian Sun, Xin Yu, Zehuan Huang, Yi-Hua Huang

Recently, methods leveraging diffusion model priors to assist monocular geometric estimation (e.g., depth and normal) have gained significant attention due to their strong generalization ability. However, most existing works focus on estimating geometric properties within the camera coordinate system of individual video frames, neglecting the inherent abilit

Mithuss Tharmalingam, Feodor Svetlanov Konomaev, Kjetil M. D. Hals

In recent years, there has been growing interest in harnessing non-collinear antiferromagnets (NCAFMs) for applications in antiferromagnetic spintronics. A key requirement for their practical use is the ability to control the spin order in a reliable and tunable manner. In this work, we investigate how the spin order in kagome antiferromagnets -- an importan

Yuqi Zhang, Yuchun Miao, Zuchao Li, Liang Ding

We introduce AMIA, a lightweight, inference-only defense for Large Vision-Language Models (LVLMs) that (1) Automatically Masks a small set of text-irrelevant image patches to disrupt adversarial perturbations, and (2) conducts joint Intention Analysis to uncover and mitigate hidden harmful intents before response generation. Without any retraining, AMIA impr

Jiatong Shi, Yifan Cheng, Bo-Hao Su, Hye-jin Shim

Speech signal analysis poses significant challenges, particularly in tasks such as speech quality evaluation and profiling, where the goal is to predict multiple perceptual and objective metrics. For instance, metrics like PESQ (Perceptual Evaluation of Speech Quality), STOI (Short-Time Objective Intelligibility), and MOS (Mean Opinion Score) each capture di

Yinqi Li, Jiahe Zhao, Hong Chang, Ruibing Hou

Contrastive Language-Image Pre-training (CLIP) has become a foundation model and has been applied to various vision and multimodal tasks. However, recent works indicate that CLIP falls short in distinguishing detailed differences in images and shows suboptimal performance on dense-prediction and vision-centric multimodal tasks. Therefore, this work focuses o

Paulo M. de Carvalho-Neto, Cícero L. Frota, Pedro G. P. Torelli

In this paper, we establish a general version of Carath\'{e}odory's existence and uniqueness theorem for a semilinear system of integro-differential equations arising from differential equations with distinct orders of Caputo fractional derivative. The main result of our work demonstrates that the integrability order of the Carath\'{e}odory function $f$ must

Harry Birch, Stéphane Régnier

The classification of solar prominences has proven to be challenging due to their diverse morphologies and dynamical behaviour. Complexity is heightened when considering eruptive prominences, where the dynamics demand methods capable of capturing detailed structural information. While there exists a range of line-of-sight (LOS) and plane-of-sky (POS) techniq

Janek Gröhl, Leonid Kunyansky, Jenni Poimala, Thomas R. Else

Quantitative comparison of the quality of photoacoustic image reconstruction algorithms remains a major challenge. No-reference image quality measures are often inadequate, but full-reference measures require access to an ideal reference image. While the ground truth is known in simulations, it is unknown in vivo, or in phantom studies, as the reference depe

Paritosh Ranjan, Surajit Majumder, Prodip Roy

Deep Learning, driven by neural networks, has led to groundbreaking advancements in Artificial Intelligence by enabling systems to learn and adapt like the human brain. These models have achieved remarkable results, particularly in data-intensive domains, supported by massive computational infrastructure. However, deploying such systems in Aerospace, where r

A. Naeimi, S. -A. Biehs

We demonstrate theoretically the non-reciprocal heating dynamics of two nanoparticles in the vicinity of a substrate all made of the ferromagnetic Weyl semi-metal Co$_3$Sn$_2$S$_2$. We show that the thermal routing effect is due to a spin-spin coupling mechanism between the nanoparticle resonances and the non-reciprocal surface modes of the substrate. Our nu

Edgar Welte, Rania Rayyes

Dexterous manipulation is a crucial yet highly complex challenge in humanoid robotics, demanding precise, adaptable, and sample-efficient learning methods. As humanoid robots are usually designed to operate in human-centric environments and interact with everyday objects, mastering dexterous manipulation is critical for real-world deployment. Traditional app

Mingyue Cheng, Jiahao Wang, Daoyu Wang, Xiaoyu Tao

Time series forecasting (TSF) is a fundamental and widely studied task, spanning methods from classical statistical approaches to modern deep learning and multimodal language modeling. Despite their effectiveness, these methods often follow a fast thinking paradigm emphasizing pattern extraction and direct value mapping, while overlooking explicit reasoning

Roberto F. Pitzalis, Nicholas Cartocci, Christian Di Natali, Darwin G. Caldwell

This paper explores the development of a control and sensor strategy for an industrial wearable wrist exoskeleton by classifying and predicting workers' actions. The study evaluates the correlation between exerted force and effort intensity, along with sensor strategy optimization, for designing purposes. Using data from six healthy subjects in a manufacturi

Binke Zhao, Ghada Alsuhi, Hani Saleh, Baker Mohammad

FALCON is a standardized quantum-resistant digital signature scheme that offers advantages over other schemes, but features more complex signature generation process. This paper presents Bi-Samplerz, a fully hardware-implemented, high-efficiency dual-path discrete Gaussian sampler designed to accelerate Falcon signature generation. Observing that the Sampler

Mostafizur Rahaman Laskar, Richa Goel

Time series forecasting is foundational in scientific and technological domains, from climate modelling to molecular dynamics. Classical approaches have significantly advanced sequential prediction, including autoregressive models and deep learning architectures such as temporal convolutional networks (TCNs) and Transformers. Yet, they remain resource-intens

Liangrui Pan, Xingchen Li, Zhongyi Chen, Ling Chu

Pathologists comprehensive evaluation of donor liver biopsies provides crucial information for accepting or discarding potential grafts. However, rapidly and accurately obtaining these assessments intraoperatively poses a significant challenge for pathologists. Features in donor liver biopsies, such as portal tract fibrosis, total steatosis, macrovesicular s

I. Tazes, S. Passalidis, G. Andrianaki, A. Skoulakis

This research demonstrates high-repetition-rate laser-accelerated ion beams via dual, intersecting, counterpropagating laser-driven blast waves to precisely shape underdense gas into long-lived near-critical density targets. The collision of the shock fronts compresses the gas and forms steep density gradients with scale lengths of a few tens of microns. The

Di Wu, Linghao Bu, Yifei Jia, Lu Cao

Recent achievements in implantable brain-computer interfaces (iBCIs) have demonstrated the potential to decode cognitive and motor behaviors with intracranial brain recordings; however, individual physiological and electrode implantation heterogeneities have constrained current approaches to neural decoding within single individuals, rendering interindividua

Nicholas Cartocci, Antonios E. Gkikakis, Roberto F. Pitzalis, Fabio Pera

Fall-caused injuries are common in all types of work environments, including offices. They are the main cause of absences longer than three days, especially for small and medium-sized businesses (SMEs). However, data, data amount, data heterogeneity, and stringent processing time constraints continue to pose challenges to real-time fall detection. This work

Eamonn Organ, Maeve Upton, Denis Allard, Lionel Benoit

Accurate high-resolution spatial and temporal wind speed data is critical for estimating the wind energy potential of a location. For real-time wind speed prediction, statistical models typically depend on high-quality (near) real-time data from official meteorological stations to improve forecasting accuracy. Personal weather stations (PWS) offer an additio

Ignacio Boero, Santiago Diaz, Tomás Vázquez, Enzo Coppes

The Optimal Reactive Power Dispatch (ORPD) problem plays a crucial role in power system operations, ensuring voltage stability and minimizing power losses. Recent advances in machine learning, particularly within the ``learning to optimize'' framework, have enabled fast and efficient approximations of ORPD solutions, typically by training models on precomput

Philipp Birken, Andreas Dedner, Robert Klöfkorn

Discontinuous Galerkin (DG) methods are promising high order discretizations for unsteady compressible flows. Here, we focus on Numerical Weather Prediction (NWP). These flows are characterized by a fine resolution in $z$-direction and low Mach numbers, making the system stiff. Thus, implicit time integration is required and for this a fast, highly parallel,

Dennis I. Martínez-Moreno, Miguel Castillo-Celeita, Diego G. Bussandri

Predicting the outcomes of quantum measurements is a cornerstone of quantum information theory and a key resource for quantum technologies. Here, we introduce a comprehensive framework for quantifying the predictability of measurements on a bipartite quantum system using error measures inherited from statistical learning theory: the Bayes risk and inference

Mehdi Moradi, Matthias Eckardt

Spatial phenomena in environmental and biological contexts often involve events that are unevenly distributed across space and carry attributes, whose associations/variations are space-dependent. In this paper, we introduce the class of inhomogeneous mark correlation functions, capturing mark associations/variations, while explicitly accounting for the spati

Guiyang Hou, Xing Gao, Yuchuan Wu, Xiang Huang

Recently, Large Language Models (LLMs) have made significant progress in IQ-related domains that require careful thinking, such as mathematics and coding. However, enhancing LLMs' cognitive development in social domains, particularly from a post-training perspective, remains underexplored. Recognizing that the social world follows a distinct timeline and req

Ximing Xing, Ziteng Xue, Yandong Guan, Jing Zhang

Generating high-quality Scalable Vector Graphics (SVGs) is challenging for Large Language Models (LLMs), as it requires advanced reasoning for structural validity, semantic accuracy, and visual coherence -- areas where current LLMs often struggle. In this work, we introduce Reason-SVG, a novel framework equipped with enhanced structured reasoning for SVG gen

Liangrui Pan, Qingchun Liang, Shen Zhao, Songqing Fan

Accurately predicting gene mutations, mutation subtypes and their exons in lung cancer is critical for personalized treatment planning and prognostic assessment. Faced with regional disparities in medical resources and the high cost of genomic assays, using artificial intelligence to infer these mutations and exon variants from routine histopathology images

Andres Fernandez, Juan Azcarreta, Cagdas Bilen, Jesus Monge Alvarez

Recent work in online speech spectrogram inversion effectively combines Deep Learning with the Gradient Theorem to predict phase derivatives directly from magnitudes. Then, phases are estimated from their derivatives via least squares, resulting in a high quality reconstruction. In this work, we introduce three innovations that drastically reduce computation

E. V. Manzhelii, S. B. Feodosyev

The characteristics of discrete vibrational levels caused by a point three-parameter substitutional impurity in long linear chain of inert gas atoms adsorbed in groove on the surface of carbon nanobundle are studied. The impurity atom differs from the atoms of the chain in the following parameters: the mass, the parameter of interaction with neighboring atom

Wenrui Liu, Qian Chen, Wen Wang, Yafeng Chen

Neural audio codecs, used as speech tokenizers, have demonstrated remarkable potential in the field of speech generation. However, to ensure high-fidelity audio reconstruction, neural audio codecs typically encode audio into long sequences of speech tokens, posing a significant challenge for downstream language models in long-context modeling. We observe tha

Sandip Kumar Maiti, Satyajit Sahoo, Gorachand Chakraborty

In this paper, we characterize the convexity of the Berezin range for finite-rank operators acting on the weighted Hardy space $\mathcal{H}_\gamma (\mathbb{D})$ over the unit disc $\mathbb{D}$. We provide a complete classification in terms of convexity for concrete operators. Additionally, we address dynamical properties of finite-rank operators on Hardy and

Xia Zhao, Peibiao Zhao

P. Salani [Adv. Math., 229 (2012)] introduced the $k$-torsional rigidity associated with a $k$-Hessian equation and obtained the Brunn-Minkowski inequalities $w.r.t.$ the torsional rigidity in $\mathbb{R}^3$. Following this work, we first construct, in the present paper, a Hadamard variational formula for the $k$-torsional rigidity with $1\leq k\leq n-1$, th

Xin Jing, Jiadong Wang, Iosif Tsangko, Andreas Triantafyllopoulos

Although speech emotion recognition (SER) has advanced significantly with deep learning, annotation remains a major hurdle. Human annotation is not only costly but also subject to inconsistencies annotators often have different preferences and may lack the necessary contextual knowledge, which can lead to varied and inaccurate labels. Meanwhile, Large Langua

David Steinmann, Wolfgang Stammer, Antonia Wüst, Kristian Kersting

Developing high-performing, yet interpretable models remains a critical challenge in modern AI. Concept-based models (CBMs) attempt to address this by extracting human-understandable concepts from a global encoding (e.g., image encoding) and then applying a linear classifier on the resulting concept activations, enabling transparent decision-making. However,

M. Kazarian, E. Krasilnikov, S. Lando, M. Shapiro

Weight systems are functions on chord diagrams satisfying Vassiliev's $4$-term relations. They originate in the theory of finite type knot invariants. Recent developments in understanding weight systems arising from Lie algebras are based on extending these weight systems from chord diagrams (which can be interpreted as involutions without fixed points, cons

Maximilian Pfister

We study the maximum number of straight-line segments connecting $n$ points in convex position in the plane, so that each segment intersects at most $k$ others. This question can also be framed as the maximum number of edges of an outer $k$-planar graph on $n$ vertices. We outline several approaches to tackle the problem with the best approach yielding an up

Anasse Boutayeb, Iyad Lahsen-cherif, Ahmed El Khadimi

Object detection has recently seen an interesting trend in terms of the most innovative research work, this task being of particular importance in the field of remote sensing, given the consistency of these images in terms of geographical coverage and the objects present. Furthermore, Deep Learning (DL) models, in particular those based on Transformers, are

J. Braß, D. M. Müller, S. Villalba-Chávez, K. Krajewska

Production of electron-positron pairs in the superposition of oscillating electric-field pulses with largely different frequencies is studied, focussing on the impact of relative phases between the pulses. Various field configurations are considered: superpositions of either two or three pulses of equal duration as well as combinations of a long low-frequenc

Nicholas Cartocci, Antonios E. Gkikakis, Darwin G. Caldwell, Jesús Ortiz

Developing a general-purpose wearable real-time fall-detection system is still a challenging task, especially for healthy and strong subjects, such as industrial workers that work in harsh environments. In this work, we present a hybrid approach for fall detection and prevention, which uses the dynamic model of an inverted pendulum to generate simulations of

Falih Gozi Febrinanto, Kristen Moore, Chandra Thapa, Jiangang Ma

The performance of existing audio deepfake detection frameworks degrades when confronted with new deepfake attacks. Rehearsal-based continual learning (CL), which updates models using a limited set of old data samples, helps preserve prior knowledge while incorporating new information. However, existing rehearsal techniques don't effectively capture the dive

Omar Benhar, Lucas Tonetto

The temperature of a newly formed neutron star is believed to be as high as $10^{11}$~K, corresponding to a thermal energy of about $10$ MeV. After a time $t \sim 50 \ {\rm s}$, the neutrino mean free path in nuclear matter exceeds the typical star radius, R~$\sim$~10 Km, and neutrino emission becomes the dominant mechanism of energy loss, eventually bringin

Mohamed Habibi, Hamza Hafsi

In this work we investigate the transfer of fundamental order and completeness properties between truncated Riesz spaces and their unitizations. Specifically, we provide characterizations and equivalences for several notions of completeness: the Archimedean property, relatively uniform completeness, Dedekind completeness, lateral completeness, universal comp

Yonathan Sarmiento, Benjamin Walter, Debraj Das, Samvit Mahapatra

We explore first-passage phenomenology for biased active processes with a renewal-type structure, focusing in particular on paradigmatic run-and-tumble models in both discrete and continuous state spaces. In general, we show there is no equality between distributions of conditional first-passage times to symmetric barriers positioned in and against the bias

Yuchong Li, Xiaojun Zeng, Chihua Fang, Jian Yang

Hepato-pancreato-biliary (HPB) disorders represent a global public health challenge due to their high morbidity and mortality. Although large language models (LLMs) have shown promising performance in general medical question-answering tasks, the current evaluation benchmarks are mostly derived from standardized examinations or manually designed questions, l

Arian Baloochestani, Leander Jehl

Blockchain protocols incentivize participation through monetary rewards, assuming rational actors behave honestly to maximize their gains. However, attackers may attempt to harm others even at personal cost. These denial of profit attacks aim to reduce the rewards of honest participants, potentially forcing them out of the system. While existing work has lar

Jing Huang, Yongkang Zhao, Yuhan Li, Zhitao Dai

The U-shaped encoder-decoder architecture with skip connections has become a prevailing paradigm in medical image segmentation due to its simplicity and effectiveness. While many recent works aim to improve this framework by designing more powerful encoders and decoders, employing advanced convolutional neural networks (CNNs) for local feature extraction, Tr

Fei Bai, Yingqian Min, Beichen Zhang, Zhipeng Chen

In this paper, we investigate code-integrated reasoning, where models generate code when necessary and integrate feedback by executing it through a code interpreter. To acquire this capability, models must learn when and how to use external code tools effectively, which is supported by tool-augmented reinforcement learning (RL) through interactive learning.

Sania Nayab, Marco Simoni, Giulio Rossolini

The rapid spread of misinformation, further amplified by recent advances in generative AI, poses significant threats to society, impacting public opinion, democratic stability, and national security. Understanding and proactively assessing these threats requires exploring methodologies that enable structured and scalable misinformation generation. In this pa

Vasilije Markovic, Lazar Obradovic, Laszlo Hajdu, Jovan Pavlovic

Integrating Large Language Models (LLMs) with Knowledge Graphs (KGs) results in complex systems with numerous hyperparameters that directly affect performance. While such systems are increasingly common in retrieval-augmented generation, the role of systematic hyperparameter optimization remains underexplored. In this paper, we study this problem in the cont

LearnLM Team, Abhinit Modi, Aditya Srikanth Veerubhotla, Aliya Rysbek

Artificial intelligence (AI) is poised to transform education, but the research community lacks a robust, general benchmark to evaluate AI models for learning. To assess state-of-the-art support for educational use cases, we ran an "arena for learning" where educators and pedagogy experts conduct blind, head-to-head, multi-turn comparisons of leading AI mode

Yuting Zhang, Hao Lu, Qingyong Hu, Yin Wang

Periodic or quasi-periodic phenomena reveal intrinsic characteristics in various natural processes, such as weather patterns, movement behaviors, traffic flows, and biological signals. Given that these phenomena span multiple modalities, the capabilities of Multimodal Large Language Models (MLLMs) offer promising potential to effectively capture and understa

Cheng Zeng, Xiatian Qi, Chi Chen, Kai Sun

Transformers have been seldom employed in point cloud roof plane instance segmentation, which is the focus of this study, and existing superpoint Transformers suffer from limited performance due to the use of low-quality superpoints. To address this challenge, we establish two criteria that high-quality superpoints for Transformers should satisfy and introdu

L. V. Lokutsievskiy, M. I. Zelikin

In this paper, we provide examples of Finsler and sub-Finsler manifolds whose geodesics exhibit chattering, that is, a countable number of switches over an arbitrarily small time interval. We also present an explicit left-invariant structure on a Carnot group whose geodesics exhibit chattering. This provides a negative answer to Le Donne's question. Furtherm

Nikita Balagansky, Yaroslav Aksenov, Daniil Laptev, Vadim Kurochkin

Sparse Autoencoders (SAEs) have proven to be powerful tools for interpreting neural networks by decomposing hidden representations into disentangled, interpretable features via sparsity constraints. However, conventional SAEs are constrained by the fixed sparsity level chosen during training; meeting different sparsity requirements therefore demands separate

Hieu Tran, Phuong-Anh Nguyen-Le, Huy Nghiem, Quang-Nhan Nguyen

Machine translation (MT) systems universally degrade when faced with code-mixed text. This problem is more acute for low-resource languages that lack dedicated parallel corpora. This work directly addresses this gap for Vietnamese-English, a language context characterized by challenges including orthographic ambiguity and the frequent omission of diacritics

Dinesh Kumar, Kapil Kumar, N. K. Karn, Ganesh Gurjar

This article reports the synthesis of a single crystalline mixed topological insulator (TI) BiSbTe$_3$ and its detailed structural and magneto-transport properties. The single crystalline samples of BiSbTe$_3$ are grown by the melt-growth process and characterized by X-ray diffraction (XRD), Energy dispersive X-ray analysis (EDAX) and Raman spectroscopy. The

Lakhan V. Jaybhaye

Over the past century, General Relativity (GR) has been a cornerstone of gravitational theory. However, recent cosmological observations, such as the accelerated expansion of the Universe, challenge its completeness and the standard $\Lambda$CDM model. This has motivated the development of alternative approaches, including dynamical dark energy and modificat

Christina Runkel, Natacha Kuete Meli, Jovita Lukasik, Ander Biguri

Compressing and pruning large machine learning models has become a critical step towards their deployment in real-world applications. Standard pruning and compression techniques are typically designed without taking the structure of the network's weights into account, limiting their effectiveness. We explore the impact of smooth regularization on neural netw

Simanta Lahkar, Valeria Bragaglia, Behnaz Bagheri, Donato Francesco Falcone

Resistive random access memories (ReRAMs) with a bilayer TaOx/HfO2 stack structure have shown unique multi-level resistive switching capabilities. However, the physical processes governing their behavior, and specifically the atomistic mechanisms of forming, remain poorly understood. In this work, we present a detailed analysis of the forming mechanism at th

Dariusz Chruściński, Frederik vom Ende, Gen Kimura, Paolo Muratore-Ginanneschi

Relaxation rates are key characteristics of quantum processes, as they determine how quickly a quantum system thermalizes, equilibrates, decoheres, and dissipates. While they play a crucial role in theoretical analyses, relaxation rates are also often directly accessible through experimental measurements. Recently, it was shown that for quantum processes gov

Yingjia Xu, Jinlin Wu, Daming Gao, Zhen Chen

Text-based person retrieval aims to identify a target individual from an image gallery using a natural language description. Existing methods primarily focus on appearance-driven cross-modal retrieval, yet face significant challenges due to the visual complexity of scenes and the inherent ambiguity of textual descriptions. The contextual information, such as

Miguel A. Sabogal, Rafael C. Nunes

Recent analyses joining data from the Cosmic Microwave Background (CMB), Baryon Acoustic Oscillations (BAO), and Type Ia Supernovae (SNIa) have provided strong evidence in favor of dynamical dark energy (DDE) over a simple cosmological constant. Motivated by these findings, we present new observational constraints on DDE based on the cross-correlation betwee

Manojlo Vukovic, Dusan Jakovetic, Dragana Bajovic, Soummya Kar

We consider a standard distributed optimization problem in which networked nodes collaboratively minimize the sum of their locally known convex costs. For this setting, we address for the first time the fundamental problem of design and analysis of distributed methods to solve the above problem when inter-node communication is subject to \emph{heavy-tailed}

Andrea Lira-Loarca, Francesco Ferrari, Andrea Mazzino

Accurate projections of wind energy potential under climate change are critical for effective long-term energy planning. While previous studies have highlighted the value of multi-model ensembles, they often fall short in capturing the full spectrum of uncertainties and temporal dynamics relevant to wind resource reliability. This paper presents one of the m

Jake Taylor, Michael Radica, Richard D. Chatterjee, Mark Hammond

We present a JWST NIRISS/SOSS transmission spectrum of the super-Earth GJ 357 b: the first atmospheric observation of this exoplanet. Despite missing the first $\sim$40 % of the transit due to using an out-of-date ephemeris, we still recover a transmission spectrum that does not display any clear signs of atmospheric features. We perform a search for Gaussia

Mahe Zabin, Ho-Jin Choi, Md. Monirul Islam, Jia Uddin

The performance of a classifier depends on the tuning of its parame ters. In this paper, we have experimented the impact of various tuning parameters on the performance of a deep convolutional neural network (DCNN). In the ex perimental evaluation, we have considered a DCNN classifier that consists of 2 convolutional layers (CL), 2 pooling layers (PL), 1 dro

Jingyao Li, Senqiao Yang, Sitong Wu, Han Shi

In recent years, developing compact and efficient large language models (LLMs) has emerged as a thriving area of research. Traditional Supervised Fine-Tuning (SFT), which relies on singular ground truth labels, often fails to capture token-level dependencies and linguistic diversity. To address these limitations, we propose a logits-based fine-tuning framewo

Falih Gozi Febrinanto, Adonia Simango, Chengpei Xu, Jingjing Zhou

Graph neural networks (GNNs) have been developed to model the relationship between regions of interest (ROIs) in brains and have shown significant improvement in detecting brain diseases. However, most of these frameworks do not consider the intrinsic relationship of causality factor between brain ROIs, which is arguably more essential to observe cause and e

Fangzhou Guo, Jibo He

Matched filtering is a common method for detecting gravitational waves. However, the computational costs of searching large template banks limit the efficiency of classical algorithms when searching for massive black hole binary (MBHB) systems. This work explores the application of a quantum matched filtering algorithm based on Grover's algorithm to MBHB sig

Tianlong Yu, Chenghang Ye, Zheyu Yang, Ziyi Zhou

The SEAR Dataset is a novel multimodal resource designed to study the emerging threat of social engineering (SE) attacks orchestrated through augmented reality (AR) and multimodal large language models (LLMs). This dataset captures 180 annotated conversations across 60 participants in simulated adversarial scenarios, including meetings, classes and networkin

Yan Bai, Elena M. Martinez, Mizuki Yamanaka, Marko Rissanen

Using real-world food price and greenhouse gas (GHG) emissions data for locally available food items in 171 countries, we measure how healthy diets could be obtained with the lowest possible emissions, compared to costs and emissions of the least expensive options and foods most commonly consumed. We find that foods with the lowest GHG emissions for a health

Emilio Villa-Cueva, Sholpan Bolatzhanova, Diana Turmakhan, Kareem Elzeky

Translating cultural content poses challenges for machine translation systems due to the differences in conceptualizations between cultures, where language alone may fail to convey sufficient context to capture region-specific meanings. In this work, we investigate whether images can act as cultural context in multimodal translation. We introduce CaMMT, a hu

Cesar Gonzalez-Gutierrez, Ariadna Quattoni

This empirical study analyzes the effects of the pre-training corpus on the quality of learned transformer representations. We focus on the representation quality induced solely through pre-training. Our experiments show that pre-training on a small, specialized corpus can yield effective representations, and that the success of combining a generic and a spe

Xi Chen, Matti Lassas, Lauri Oksanen, Gabriel P. Paternain

We pose and solve an inverse problem for the classical field equations that arise in the Standard Model of particle physics. Our main result describes natural conditions on the representations, so that it is possible to recover all the fields from measurements in a small set within a causal domain in Minkowski space. These conditions are satisfied for the re

Anda Tang, Yiming Dong, Yutao Zeng, zhou Xun

The expanding computational costs and limited resources underscore the critical need for budgeted-iteration training, which aims to achieve optimal learning within predetermined iteration budgets. While learning rate schedules fundamentally govern the performance of different networks and tasks, particularly in budgeted-iteration scenarios, their design rema