March 2025 arXiv papers — page 180

Koji Inoue, Yuki Okafuji, Jun Baba, Yoshiki Ohira

Turn-taking is a crucial aspect of human-robot interaction, directly influencing conversational fluidity and user engagement. While previous research has explored turn-taking models in controlled environments, their robustness in real-world settings remains underexplored. In this study, we propose a noise-robust voice activity projection (VAP) model, based o

Isaac Dobes, Naihuan Jing

In this paper we present a modified version of the proof given Jing-Yang-Zhao's paper "Local Unitary Equivalence of Quantum States and Simultaneous Orthogonal Equivalence," which established the correspondence between local unitary (LU) equivalence and simultaneous orthogonal equivalence of bipartite quantum states. Our modified proof utilizes a hypermatrix

Yongtian Luo, Amrinder S. Nain, Brian A. Camley

Interactions between crawling cells, which are essential for many biological processes, can be quantified by measuring cell-cell collisions. Conventionally, experiments of cell-cell collisions are conducted on two-dimensional flat substrates, where colliding cells repolarize and move away upon contact with one another in "contact inhibition of locomotion" (C

Kibum Kim, Sein Kim, Hongseok Kang, Jiwan Kim

Large Language Models (LLMs) have recently emerged as a powerful backbone for recommender systems. Existing LLM-based recommender systems take two different approaches for representing items in natural language, i.e., Attribute-based Representation and Description-based Representation. In this work, we aim to address the trade-off between efficiency and effe

Yifan Chang, Junjie Huang, Xiaofeng Wang, Yun Ye

Monocular 3D lane detection is a fundamental task in autonomous driving. Although sparse-point methods lower computational load and maintain high accuracy in complex lane geometries, current methods fail to fully leverage the geometric structure of lanes in both lane geometry representations and model design. In lane geometry representations, we present a th

Zhaori Liu, Mengyang Li, Hu Han, Enli Zhang

Medical image segmentation is essential for clinical diagnosis, surgical planning, and treatment monitoring. Traditional approaches typically strive to tackle all medical image segmentation scenarios via one-time learning. However, in practical applications, the diversity of scenarios and tasks in medical image segmentation continues to expand, necessitating

Yang LI, Jinglu Wang, Lei Chu, Xiao Li

The advent of 3D Gaussian Splatting (3DGS) has advanced 3D scene reconstruction and novel view synthesis. With the growing interest of interactive applications that need immediate feedback, online 3DGS reconstruction in real-time is in high demand. However, none of existing methods yet meet the demand due to three main challenges: the absence of predetermine

Eric A. Bergshoeff, Ergin Sezgin, Paul K. Townsend

``When to the sessions of sweet silent thought I summon up re-membranes of things past, I sigh the lack of many a thing I sought''. (Apologies to William Shakespeare)

Amana Liaqat, Ahmed Darwish, Adrian Roman, Stephen DiAdamo

Scaling quantum computing requires networked systems, leveraging HPC for distributed simulation now and quantum networks in the future. Quantum datacenters will be the primary access point for users, but current approaches demand extensive manual decisions and hardware expertise. Tasks like algorithm partitioning, job batching, and resource allocation divert

Yanjun Chen, Yirong Sun, Xinghao Chen, Jian Wang

Chain-of-Thought (CoT) reasoning has proven effective in natural language tasks but remains underexplored in multimodal alignment. This study investigates its integration into 3D vision-language learning by embedding structured reasoning into alignment training. We introduce the 3D-CoT Benchmark, a dataset with hierarchical CoT annotations covering shape rec

Aditya Shankar, Lydia Y. Chen, Arie van Deursen, Rihan Hai

Generating temporal data under conditions is crucial for forecasting, imputation, and generative tasks. Such data often has metadata and partially observed signals that jointly influence the generated values. However, existing methods face three key limitations: (1) they condition on either the metadata or observed values, but rarely both together; (2) they

Samuel Zamour

In line with known results on groups, we show the existence of the nilpotent radical in Lie rings with minimal condition on centralizers. We also prove a form of Engel's theorem if the characteristic is zero.

Federico Mazzoni, Riccardo Guidotti, Alessio Malizia

We introduce Frank, a human-in-the-loop system for co-evolutionary hybrid decision-making aiding the user to label records from an un-labeled dataset. Frank employs incremental learning to ``evolve'' in parallel with the user's decisions, by training an interpretable machine learning model on the records labeled by the user. Furthermore, Frank advances state

Hemjyoti Nath, Abhishek Sarma

In this paper, we consider the set of partitions $ped(n)$ which counts the number of partitions of $n$ wherein the even parts are distinct (and the odd parts are unrestricted). Using an algorithm developed by Radu, we prove congruences modulo 192 which were conjectured by Nath. Further, we prove a few infinite families of congruences modulo 24 by using a res

Ruixiang Wang, Huayi Zhou, Xinyue Yao, Guiliang Liu

Achieving precise and generalizable grasping across diverse objects and environments is essential for intelligent and collaborative robotic systems. However, existing approaches often struggle with ambiguous affordance reasoning and limited adaptability to unseen objects, leading to suboptimal grasp execution. In this work, we propose GAT-Grasp, a gesture-dr

Kedi Xie, Martin Guay, Shimin Wang, Fang Deng

This paper studies the linear quadratic regulation (LQR) problem of unknown discrete-time systems via dynamic output feedback learning control. In contrast to the state feedback, the optimality of the dynamic output feedback control for solving the LQR problem requires an implicit condition on the convergence of the state observer. Moreover, due to unknown s

Feng Sun, Xuqiang Zhang, Jiangtao Chen, Yun Zhao

Single-atom catalysts (SACs) have attracted ever-growing interest due to their high atom-utilization efficiency and potential for cost-effective of hydrogen production. However, enhancing the hydrogen evolution reaction (HER) performance remains a key challenge in developing SACs for HER technology. Herein, we employed first-principles calculations in conjun

Edgar Assing, Radu Toma

The orbit method in its quantitative form due to Nelson and Venkatesh has played a central role in recent advances in the analytic theory of higher rank $L$-functions. The goal of this note is to explain how the method can be applied to the sup-norm problem for automorphic forms on $\operatorname{PGL}(2)$. Doing so, we prove a new hybrid bound for newforms $

Meng Wang, Fan Wu, Yunchuan Qin, Ruihui Li

The vision-based semantic scene completion task aims to predict dense geometric and semantic 3D scene representations from 2D images. However, the presence of dynamic objects in the scene seriously affects the accuracy of the model inferring 3D structures from 2D images. Existing methods simply stack multiple frames of image input to increase dense scene sem

Saikat Panja, Prachi Saini, Anupam Singh

Let $\mathbf{O}(\mathbb{F})$ be the split octonion algebra over an algebraically closed field $\mathbb{F}$. For positive integers $k_1, k_2\geq 2$, we study surjectivity of the map $A_1(x^{k_1}) + A_2(y^{k_2}) \in \mathbf{O}(\mathbb{F})\langle x, y\rangle$ on $\mathbf{O}(\mathbb{F})$. For this, we use the orbit representatives of the ${G}_2(\mathbb{F})$-acti

Xin Ding, Hao Wu, Yifan Yang, Shiqi Jiang

With the rise of real-world human-AI interaction applications, such as AI assistants, the need for Streaming Video Dialogue is critical. To address this need, we introduce StreamMind, a video LLM framework that achieves ultra-FPS streaming video processing (100 fps on a single A100) and enables proactive, always-on responses in real time, without explicit us

Meng Wang, Huilong Pi, Ruihui Li, Yunchuan Qin

Camera-based 3D semantic scene completion (SSC) provides dense geometric and semantic perception for autonomous driving. However, images provide limited information making the model susceptible to geometric ambiguity caused by occlusion and perspective distortion. Existing methods often lack explicit semantic modeling between objects, limiting their percepti

Weidong Zhan, Yue Wang, Nan Hu, Liming Xiao

Currently, long-chain reasoning remains a key challenge for large language models (LLMs) because natural texts lack sufficient explicit reasoning data. However, existing benchmarks suffer from limitations such as narrow coverage, short reasoning paths, or high construction costs. We introduce SCoRE (Scenario-based Commonsense Reasoning Evaluation), a benchma

Daniel L. Rodríguez-Vidanes, Juan Carlos Sampedro

We investigate the isometric structure of $L^{p}$-spaces for the infinite-dimensional Lebesgue measure $(\mathbb{R}^{\mathbb{N}},\mu)$. Under the continuum hypothesis (CH) we prove $L^{p}(\mu)\cong \ell^{p}(\mathfrak{c},L^{p}[0,1])$, where $\mathfrak{c}$ denotes the cardinality of the continuum, and without CH we obtain an isometric, complemented copy of $\e

Huapeng Lin, Miao Yu

Distributed photovoltaic (DPV) systems are essential for advancing renewable energy applications and achieving energy independence. Accurate DPV power forecasting can optimize power system planning and scheduling while significantly reducing energy loss, thus enhancing overall system efficiency and reliability. However, solar energy's intermittent nature and

Jiawei Chen, Jinsong Yang

Recently, two new quantum-corrected black hole models satisfying covariance have been proposed within the framework of effective quantum gravity. In this paper, we study how the quantum parameter $\zeta$ affects the optical properties of two quantum-corrected black hole models. We first analyze the photon sphere, critical impact parameter, and innermost stab

Ruiyu Wang, Sen Wang, Xinxin Zuo, Qiang Sun

Lifelong learning (LL) aims to continuously acquire new knowledge while retaining previously learned knowledge. A central challenge in LL is the stability-plasticity dilemma, which requires models to balance the preservation of previous knowledge (stability) with the ability to learn new tasks (plasticity). While parameter-efficient fine-tuning (PEFT) has be

Yue Jin, Yongchao Liu, Chuntao Hong

Graph-based computations are crucial in a wide range of applications, where graphs can scale to trillions of edges. To enable efficient training on such large graphs, mini-batch subgraph sampling is commonly used, which allows training without loading the entire graph into memory. However, existing solutions face significant trade-offs: online subgraph gener

Santiago Cuervo, Adel Moumen, Yanis Labrak, Sameer Khurana

Text-pretrained language models (LMs) encode rich world knowledge, but adapting them to process and generate perceptual modalities such as audio and images while effectively leveraging that knowledge remains challenging. Perceptual modalities are finer-grained and less semantically dense than text, making it unclear how functions learned during text pretrain

Jeffery Ezearn

Let $\chi$ be a non-principal Dirichlet character of modulus $q$ with associated \textit{L}-function $L(s,\chi)$. We prove that $$|L(1,\chi)|\le\left(\frac{1}{2}+O\Big(\frac{\log\log q}{\log q}\Big)\right)\frac{\varphi(q)}{q}\log q\,,$$ where $\varphi(\cdot)$ is Euler's phi function. This refines known bounds of the form $(c+o(1))\log q $ or $(c+O(\frac{1}{\

Camelia Jana, Monu Singh, Suvendu Rakshit, Santabrata Das

We examine mass outflows from a low-angular momentum, viscous, advective, and magnetized accretion disk around a rotating black hole in presence of thermal conduction. We consider the disk is primarily threaded by the toroidal component of the magnetic field and an effective potential satisfactorily mimicked the spacetime geometry around the rotating black h

Khoi Do, Binh-Son Hua

Score distillation sampling is an effective technique to generate 3D models from text prompts, utilizing pre-trained large-scale text-to-image diffusion models as guidance. However, the produced 3D assets tend to be over-saturating, over-smoothing, with limited diversity. These issues are results from a reverse Kullback-Leibler (KL) divergence objective, whi

Xiabao Wu, Yongchao Liu, Wei Qin, Chuntao Hong

Graph neural networks (GNNs) have delivered remarkable results in various fields. However, the rapid increase in the scale of graph data has introduced significant performance bottlenecks for GNN inference. Both computational complexity and memory usage have risen dramatically, with memory becoming a critical limitation. Although graph sampling-based subgrap

I. Kuntz, R. da Rocha

A functional measure encompasses quantum corrections and is explored in the fluid/gravity correspondence. Corrections to response and transport coefficients in the second-order dissipative relativistic hydrodynamics are proposed, including the ones to the pressure, the relaxation time, and the shear relaxation. Their dependence on the quark-gluon plasma (QGP

P. C. da Silva Junior, O. P. Ferreira, G. N. Silva

This paper presents a novel variant of the Broyden quasi-Newton secant-type method aimed at solving constrained mixed generalized equations, which can include functions that are not necessarily differentiable. The proposed method integrates the classical secant approach with techniques inspired by the Conditional Gradient method to handle constraints effecti

Manuel Cañizares, Pedro Caro, Ioannis Parissis, Thanasis Zacharopoulos

The initial-to-final-state inverse problem consists in determining a quantum Hamiltonian assuming the knowledge of the state of the system at some fixed time, for every initial state. This problem was formulated by Caro and Ruiz and motivated by the data-driven prediction problem in quantum mechanics. Caro and Ruiz analysed the question of uniqueness for Ham

Oriel Perets, Ofir Ben Shoham, Nir Grinberg, Nadav Rappoport

Medical benchmark datasets significantly contribute to developing Large Language Models (LLMs) for medical knowledge extraction, diagnosis, summarization, and other uses. Yet, current benchmarks are mainly derived from exam questions given to medical students or cases described in the medical literature, lacking the complexity of real-world patient cases tha

Nikumbh Sarthak Sham, Sandip Chakraborty, Shamik Sural

While a plethora of machine learning (ML) models are currently available, along with their implementation on disparate platforms, there is hardly any verifiable ML code which can be executed on public blockchains. We propose a novel approach named LMST that enables conversion of the inferencing path of an ML model as well as its weights trained off-chain int

Wei Liu, Zhiying Deng, Zhongyu Niu, Jun Wang

Extracting a small subset of crucial rationales from the full input is a key problem in explainability research. The most widely used fundamental criterion for rationale extraction is the maximum mutual information (MMI) criterion. In this paper, we first demonstrate that MMI suffers from diminishing marginal returns. Once part of the rationale has been iden

Yixin Wu, Feiran Zhang, Tianyuan Shi, Ruicheng Yin

Recent advances in diffusion models have enabled the creation of deceptively real images, posing significant security risks when misused. In this study, we empirically show that different timesteps of DDIM inversion reveal varying subtle distinctions between synthetic and real images that are extractable for detection, in the forms of such as Fourier power s

Yecong Wan, Mingwen Shao, Yuanshuo Cheng, Jun Shu

Videos captured under real-world adverse weather conditions typically suffer from uncertain hybrid weather artifacts with heterogeneous degradation distributions. However, existing algorithms only excel at specific single degradation distributions due to limited adaption capacity and have to deal with different weather degradations with separately trained mo

Xinru Wang, Tanujit Chakraborty, Bibhas Chakraborty

Dynamic treatment regimes (DTRs) are sequences of decision rules designed to tailor treatment based on patients' treatment history and evolving disease status. Ordinal outcomes frequently serve as primary endpoints in clinical trials and observational studies. However, constructing optimal DTRs for ordinal outcomes has been underexplored. This paper introduc

Em K. Thompson

We describe five ideal triangulations of the 3-cusped hyperbolic `magic manifold' that are each compatible with well-established techniques for triangulating Dehn fillings. Using these techniques, we construct low-complexity triangulations for all partial fillings of the magic manifold, and in particular, recover minimal triangulations for 229 of the hyperbo

Yaswanth Kumar LS, Somya Jain, Bheemarjuna Reddy Tamma, Koteswararao Kondepu

O-RAN has brought in deployment flexibility and intelligent RAN control for mobile operators through its disaggregated and modular architecture using open interfaces. However, this disaggregation introduces complexities in system integration and network management, as components are often sourced from different vendors. In addition, the operators who are rel

Shashata Sawmya, Thomas L. Athey, Gwyneth Liu, Nir Shavit

Training segmentation models from scratch has been the standard approach for new electron microscopy connectomics datasets. However, leveraging pretrained models from existing datasets could improve efficiency and performance in constrained annotation budget. In this study, we investigate domain adaptation in connectomics by analyzing six major datasets span

Agnia Sergeyuk, Ilya Zakharov, Ekaterina Koshchenko, Maliheh Izadi

The integration of Artificial Intelligence (AI) into Integrated Development Environments (IDEs) is reshaping software development, fundamentally altering how developers interact with their tools. This shift marks the emergence of Human-AI Experience in Integrated Development Environment (in-IDE HAX), a field that explores the evolving dynamics of Human-Compu

Hyuga Yoshizaki

Let $p$ be a prime number. The $p$-power cyclic resultant of a polynomial is the determinant of the Sylvester matrix of $t^{p^n}-1$ and the polynomial. It is known that the sequence of $p$-power cyclic resultants and its non-$p$-parts converge in $\mathbb{Z}_p$. This article shows the $p$-adic convergence of the iterated $p$-power cyclic resultants of multiv

Julio Cesar Leite

A magic medallion is central in Michael Ender novel, and it is depicted as two snakes biting each other, in a loop. Folk tale says that the design of the medallion changed for the Wolfgang Petersen movie, depicting an even deeper image of infinity. The medallion turned out to be an icon for the story fans. This paper will unleash a broad view of the realm of

Luca Battaglia, Giusi Vaira, Yixing Pu

We consider the classical geometric problem of prescribing the scalar and the boundary mean curvature in the unit ball endowed with the standard Euclidean metric. We will deal with the case of negative scalar curvature showing the existence of infinitely many non-radial positive solutions when the dimension is larger or equal to 5. This is the first result o

Julián Haddad, Dylan Langharst, Galyna V. Livshyts, Eli Putterman

In 1970, Schneider introduced the $m$th-order extension of the difference body $DK$ of a convex body $K\subset\mathbb R^n$, the convex body $D^m(K)$ in $\mathbb R^{nm}$. He conjectured that its volume is minimized for ellipsoids when the volume of $K$ is fixed. In this work, we solve a dual version of this problem: we show that the volume of the polar body o

YingLiang Ma, Sandra Howell, Aldo Rinaldi, Tarv Dhanjal

Objective: Interventional devices, catheters and insertable imaging devices such as transesophageal echo (TOE) probes are routinely used in minimally invasive cardiovascular procedures. Detecting their positions and orientations in X-ray fluoroscopic images is important for many clinical applications. Method: In this paper, a novel attention mechanism was de

L. R. Colaço

This paper presents a new model-independent constraint on the Hubble constant ($H_0$) by anchoring relative distances from Type Ia supernovae (SNe Ia) observations to absolute distance measurements from time-delay strong Gravitational Lensing (SGL) systems. The approach only uses the validity of the cosmic distance duality relation (CDDR) to derive constrain

Daryna Oliynyk, Rudolf Mayer, Andreas Rauber

Machine learning models were shown to be vulnerable to model stealing attacks, which lead to intellectual property infringement. Among other methods, substitute model training is an all-encompassing attack applicable to any machine learning model whose behaviour can be approximated from input-output queries. Whereas prior works mainly focused on improving th

Si Zhou, Yain-Whar Si, Xiaochen Yuan, Xiaofan Li

In Neural Networks, there are various methods of feature fusion. Different strategies can significantly affect the effectiveness of feature representation, consequently influencing the ability of model to extract representative and discriminative features. In the field of face recognition, traditional feature fusion methods include feature concatenation and

Xiang Gao, Shuai Yang, Jiaying Liu

Optical illusion hidden picture is an interesting visual perceptual phenomenon where an image is cleverly integrated into another picture in a way that is not immediately obvious to the viewer. Established on the off-the-shelf text-to-image (T2I) diffusion model, we propose a novel training-free text-guided image-to-image (I2I) translation framework dubbed a

Xin Xu

The mean-variance (MV) model is the core of modern portfolio theory. Nevertheless, it suffers from the over-fitting problem due to the estimation errors of model parameters. We consider the $\ell_{1}$ regularized MV model, which adds an $\ell_{1}$ regularization term in the objective to prevent over-fitting and promote sparsity of solutions. By investigating

Jun Kong, Xinge Ma, Jin Wang, Xuejie Zhang

Large language models (LLMs) have achieved outstanding performance in natural language processing, but enormous model sizes and high computational costs limit their practical deployment. Structured pruning can effectively reduce the resource demands for deployment by removing redundant model parameters. However, the randomly selected calibration data and fix

Francesco Daghero, Daniele Jahier Pagliari, Francesco Conti, Luca Benini

The acceleration of pruned Deep Neural Networks (DNNs) on edge devices such as Microcontrollers (MCUs) is a challenging task, given the tight area- and power-constraints of these devices. In this work, we propose a three-fold contribution to address this problem. First, we design a set of optimized software kernels for N:M pruned layers, targeting ultra-low-

Antonio Alliegro, Francesca Pistilli, Tatiana Tommasi, Giuseppe Averta

Forecasting human-environment interactions in daily activities is challenging due to the high variability of human behavior. While predicting directly from videos is possible, it is limited by confounding factors like irrelevant objects or background noise that do not contribute to the interaction. A promising alternative is using Scene Graphs (SGs) to track

Devon Jarvis, Richard Klein, Benjamin Rosman, Andrew M. Saxe

In spite of finite dimension ReLU neural networks being a consistent factor behind recent deep learning successes, a theory of feature learning in these models remains elusive. Currently, insightful theories still rely on assumptions including the linearity of the network computations, unstructured input data and architectural constraints such as infinite wi

Felix Nagel

Multiplicative convolution $\mu \ast \nu$ of two finite signed measures $\mu$ and $\nu$ on $\mathbb{R}^n$ and a related product $\mu \circledast \nu$ on the sphere $S^{n-1}$ are studied. For fixed $\mu$ the injectivity in $\nu$ of both operations is characterised given an arbitrary group of reflections along the coordinate axes. The results for the sphere yi

Wongi Park, Myeongseok Nam, Siwon Kim, Sangwoo Jo

Recently, 3D Gaussian Splatting (3D-GS) has emerged, showing real-time rendering speeds and high-quality results in static scenes. Although 3D-GS shows effectiveness in static scenes, their performance significantly degrades in real-world environments due to transient objects, lighting variations, and diverse levels of occlusion. To tackle this, existing met

Usman Sanusi, Sona John, Johannes Mueller, Aurélien Tellier

Parasite quiescence is the ability for the pathogen to be inactive, with respect to metabolism and infectiousness, for some amount of time and then become active (infectious) again. The population is thus composed of an inactive proportion, and an active part in which evolution and reproduction takes place. In this paper, we investigate the effect of parasit

Roland Szatmári, Akio Nakahara, So Kitsunezaki, Ferenc Kun

We investigate the shrinkage induced breakup of thin layers of heterogeneous materials attached to a substrate, a ubiquitous natural phenomenon with a wide range of potential applications. Focusing on the evolution of the fragment ensemble, we demonstrate that the system has two distinct phases: damage phase, where the layer is cracked, however, a dominant p

Wenyin Wei, Liang Liao, Alexander Knieps, Jiankun Hua

In strictly axisymmetric configurations of tokamaks, field-line tracing reduces from a three-dimensional ODE system to a two-dimensional one, where Poincar\'e-Bendixson theorem applies and guarantees the nonexistence of chaos. The formulae of functional perturbation theory (FPT) mostly simplify to compact closed-form expressions to allow the computation to f

Abdullah M. Zyarah, Dhireesha Kudithipudi

The increasing demand for continual learning in sequential data processing has led to progressively complex training methodologies and larger recurrent network architectures. Consequently, this has widened the knowledge gap between continual learning with recurrent neural networks (RNNs) and their ability to operate on devices with limited memory and compute

Chang Liu, Leona Licht, Christina Wichmann, Wilhelm Eschen

Table-top extreme ultraviolet (EUV) ptychography enables nanoscale, label-free, and quantitative imaging with intrinsic elemental sensitivity, offering a unique modality for subcellular profiling of bacterial morphology and composition. In this work, we apply a state-of-the-art EUV ptychographic microscope to systematically investigate the structural and com

Soheila Emamyari, Jalal Sarabadani, Ralf Metzler, Tapio Ala-Nissila

We consider the dynamics of pore-driven polymer translocation through a nanopore to semi-infinite space when the chain is initially confined and equilibrated in a narrow channel. To this end, we use Langevin dynamics (LD) simulations and iso-flux tension propagation (IFTP) theory to characterize local and global dynamics of the translocating chain. The dynam

Jinshui Zhang, Angel Peterchev, Stefan Goetz

Noninvasive brain stimulation can activate neurons in the brain but requires power electronics with exceptionally high power in the mega-volt-ampere and high frequencies in the kilohertz range. Whereas oscillator circuits offered only one or very few pulse shapes, modular power electronics solved a long-standing problem for the first time and enabled arbitra

Shivanshu Shekhar, Tong Zhang

Diffusion models have revolutionized generative modeling in continuous domains like image, audio, and video synthesis. However, their iterative sampling process leads to slow generation and inefficient training, challenges that are further exacerbated when incorporating Reinforcement Learning from Human Feedback (RLHF) due to sparse rewards and long time hor

Youngjoon Jeong, Junha Chun, Soonwoo Cha, Taesup Kim

A world model is essential for an agent to predict the future and plan in domains such as autonomous driving and robotics. To achieve this, recent advancements have focused on video generation, which has gained significant attention due to the impressive success of diffusion models. However, these models require substantial computational resources. To addres

Shawn Li, Jiashu Qu, Yuxiao Zhou, Yuehan Qin

Vision-Language Models (VLMs) have advanced multi-modal tasks like image captioning, visual question answering, and reasoning. However, they often generate hallucinated outputs inconsistent with the visual context or prompt, limiting reliability in critical applications like autonomous driving and medical imaging. Existing studies link hallucination to stati

Mohammadreza Doostmohammadian, Zulfiya R. Gabidullina, Hamid R. Rabiee

This paper proposes an accelerated consensus-based distributed iterative algorithm for resource allocation and scheduling. The proposed gradient-tracking algorithm introduces an auxiliary variable to add momentum towards the optimal state. We prove that this solution is all-time feasible, implying that the coupling constraint always holds along the algorithm

Shawn Li, Peilin Cai, Yuxiao Zhou, Zhiyu Ni

Out-of-Distribution (OOD) detection is critical for ensuring the reliability of machine learning models in safety-critical applications such as autonomous driving and medical diagnosis. While deploying personalized OOD detection directly on edge devices is desirable, it remains challenging due to large model sizes and the computational infeasibility of on-de

Ying-Fei Zhang, Zhi-Fan Zhang, Zhen-Gang Zhu, Gang Su

In recent years, the nonlinear anomalous thermal Hall effect has attracted substantial attention. In this paper, we carry out a theoretical exploration of the intrinsic anomalous thermal Hall and Nernst effect that is induced by the thermal Berry connection polarizability. This effect is independent of the relaxation time and can be present in antiferromagne

Muhammad Umar Farooq Qaisar, Weijie Yuan, Guangjie Han, Adeel Ahmed

Wireless rechargeable sensor networks (WRSNs), supported by recent advancements in wireless power transfer (WPT) technology, hold significant potential for extending network lifetime. However, traditional approaches often prioritize scheduling algorithms and network optimization, overlooking the security risks associated with the charging process, which expo

Haotong Yang, Qingyuan Zheng, Yunjian Gao, Yongkun Yang

With the rapid advancement of text-conditioned Video Generation Models (VGMs), the quality of generated videos has significantly improved, bringing these models closer to functioning as ``*world simulators*'' and making real-world-level video generation more accessible and cost-effective. However, the generated videos often contain factual inaccuracies and l

Frédéric Dabrowski, Jordan Ischard

We introduce a functional reactive programming language that extends WORMHOLES, an enhancement of YAMPA with support for effects. Our proposal relaxes the constraint in WORMHOLES that restricts all resources to single-use. Resources are categorized into two kinds: input/output resources and internal resources. Input/output resources model interactions with t

Kai Li, Junhao Wang, William Han, Ding Zhao

Minimally invasive surgery (MIS) requires high-fidelity, real-time visual feedback of dynamic and low-texture surgical scenes. To address these requirements, we introduce FeatureEndo-4DGS (FE-4DGS), the first real time pipeline leveraging feature-distilled 4D Gaussian Splatting for simultaneous reconstruction and semantic segmentation of deformable surgical

M. Mehraeen

We present a quantum response approach to momentum-space gravity in dissipative multiband systems, which dresses both the quantum geometry--through an interband Weyl transformation--and the equations of motion. In addition to clarifying the roles of the contorsion and symplectic terms, we introduce the three-state quantum geometric tensor as a necessary elem

Davi Rodrigues, Alejandro Riveros, Andrea Meo, Emily Darwin

Skyrmions, topologically protected textures, have been observed in different fields of nanotechnology and have emerged as promising candidates for different applications due to their topological stability, low-power operation, and dynamic response to external stimuli. First introduced in particle physics, skyrmions have since been observed in different conde

Ziruo Hao, Zhenhua Cui, Tao Yang, Bo Hu

This paper provides an invariant federated learning system for resource-constrained edge intelligence. This framework can mitigate the impact of heterogeneity and asynchrony via exit strategy and invariant penalty. We introduce parameter orthogonality into edge intelligence to measure the contribution or impact of heterogeneous and asynchronous clients. It i

Baining Zhao, Jianjie Fang, Zichao Dai, Ziyou Wang

Large multimodal models exhibit remarkable intelligence, yet their embodied cognitive abilities during motion in open-ended urban 3D space remain to be explored. We introduce a benchmark to evaluate whether video-large language models (Video-LLMs) can naturally process continuous first-person visual observations like humans, enabling recall, perception, reas

Houssam Zenati, Judith Abécassis, Julie Josse, Bertrand Thirion

Uncovering causal mediation effects is of significant value to practitioners seeking to isolate the direct treatment effect from the potential mediated effect. We propose a double machine learning (DML) algorithm for mediation analysis that supports continuous treatments. To estimate the target mediated response curve, our method uses a kernel-based doubly r

Lavanya Ratnabala, Robinroy Peter, Aleksey Fedoseev, Dzmitry Tsetserukou

This paper tackles decentralized continuous task allocation in heterogeneous multi-agent systems. We present a novel framework HIPPO-MAT that integrates graph neural networks (GNN) employing a GraphSAGE architecture to compute independent embeddings on each agent with an Independent Proximal Policy Optimization (IPPO) approach for multi-agent deep reinforcem

Sabah Al-Fedaghi

This paper deals with the issue of conceptual models role in capturing semantics and aligning them to serve the remaining development phases of systems design. Specifically, the entity-relationship (ER) model is selected as an example of conceptual representation that serves this purpose in building relational database systems. It is claimed that ER diagrams

Zidu Wang, Jiankuo Zhao, Miao Xu, Xiangyu Zhu

3D Morphable Models (3DMMs) have played a pivotal role as a fundamental representation or initialization for 3D avatar animation and reconstruction. However, extending 3DMMs to hair remains challenging due to the difficulty of enforcing vertex-level consistent semantic meaning across hair shapes. This paper introduces a novel method, Semantic-consistent Ray

Fei Si, Feinuo Zhang

Motivated by asymptotic phenomena of moduli spaces of higher rank stable sheaves on algebraic surfaces, we study the Picard number of the moduli space of one-dimensional stable sheaves supported in a sufficiently positive divisor class on a surface. We give an asymptotic lower bound of the Picard number in general. In some special cases, we show that this lo

Bingyu Cui

In classical statistical mechanics, the partition function is defined in phase space. We extend this concept to quantum statistical mechanics using Bohmian trajectories. The quantum partition function in phase space captures the ensemble of positions and momenta, along with the probability distribution that accounts for the inherent uncertainty in measuring

Zixi Kang, Xinghan Wang, Yadong Mu

Human pose, action, and motion generation are critical for applications in digital humans, character animation, and humanoid robotics. However, many existing methods struggle to produce physically plausible movements that are consistent with biomechanical principles. Although recent autoregressive and diffusion models deliver impressive visual quality, they

Huan Tian, Guangsheng Zhang, Bo Liu, Tianqing Zhu

While in-processing fairness approaches show promise in mitigating biased predictions, their potential impact on privacy leakage remains under-explored. We aim to address this gap by assessing the privacy risks of fairness-enhanced binary classifiers via membership inference attacks (MIAs) and attribute inference attacks (AIAs). Surprisingly, our results rev

Jiachen Luo, Huy Phan, Lin Wang, Joshua D. Reiss

Multi-modal emotion recognition is challenging due to the difficulty of extracting features that capture subtle emotional differences. Understanding multi-modal interactions and connections is key to building effective bimodal speech emotion recognition systems. In this work, we propose Bimodal Connection Attention Fusion (BCAF) method, which includes three

Xudong Wang, Jiacheng Wang, Lei Feng, Dusit Niyato

Generative AI (GenAI) is driving the intelligence of wireless communications. Due to data limitations, random generation, and dynamic environments, GenAI may generate channel information or optimization strategies that violate physical laws or deviate from actual real-world requirements. We refer to this phenomenon as wireless hallucination, which results in

Jameel-Un Nabi, Muhammad Riaz

We present calculation of electron capture cross sections (ECC), in the limit of zero momentum transfer, using the pn QRPA model in stellar matter. Towards this aim we make use of our recently introduced recipe for estimation of nuclear partition functions. For low momentum transfer q tends to zero, the nuclear matrix elements of the P{\sigma}{\tau} plus ope

Yikun Ren, Feixiang Xu, Ming Lin

When the center of fluctuations, i.e., the nonequilibrium eigenphase, undergoes transformation, there emerge critical parameters that demonstrate insensitivity to fluctuation perturbations and even independence from the molecular physical properties of the system, while exhibiting pronounced efficacy in governing phase transition dynamics.In the context of p

Ziyue Huang, Yongchao Feng, Shuai Yang, Ziqi Liu

Remote sensing object detection has made significant progress, but most studies still focus on closed-set detection, limiting generalization across diverse datasets. Open-vocabulary object detection (OVD) provides a solution by leveraging multimodal associations between text prompts and visual features. However, existing OVD methods for remote sensing (RS) i

Xiaohong Yang, Minghui Liwang, Liqun Fu, Yuhan Su

Hierarchical Federated Learning (HFL) extends conventional Federated Learning (FL) by introducing intermediate aggregation layers, enabling distributed learning in geographically dispersed environments, particularly relevant for smart IoT systems, such as remote monitoring and battlefield operations, where cellular connectivity is limited. In these scenarios

Miquel Garcia-Fernandez

A fundamental limitation of traditional Neural Networks (NN) in predictive modelling is their inability to quantify uncertainty in their outputs. In critical applications like positioning systems, understanding the reliability of predictions is critical for constructing confidence intervals, early warning systems, and effectively propagating results. For ins

Michael Orlitzky, Giovanni Barbarino

The Lyapunov rank of a cone is the dimension of the Lie algebra of its automorphism group. It is invariant under linear isomorphism and in general not unique - two or more non-isomorphic cones can share the same Lyapunov rank. It is therefore not possible in general to identify cones using Lyapunov rank. But suppose we look only among symmetric cones. Are th

Xinan He, Yue Zhou, Bing Fan, Bin Li

Faces synthesized by diffusion models (DMs) with high-quality and controllable attributes pose a significant challenge for Deepfake detection. Most state-of-the-art detectors only yield a binary decision, incapable of forgery localization, attribution of forgery methods, and providing analysis on the cause of forgeries. In this work, we integrate Multimodal