papers

Publications (149)

cs.LG2024

RW-NSGCN: A Robust Approach to Structural Attacks via Negative Sampling

Shuqi He, Jun Zhuang, Ding Wang +1

Node classification using Graph Neural Networks (GNNs) has been widely applied in various practical scenarios, such as predicting user interests and detecting communities in social…

cs.CV2026

AV-SyncBench: Decoupled Benchmarking of Temporal and Semantic Audio-Visual Synchronization

Tianhong Zhou, Mingyang Han, Boyu Li +8

Audio-visual feature extraction is a fundamental component of multimodal understanding and generation tasks. However, existing evaluation protocols for feature extraction models ex…

hep-ph2017

New insights into hadron production mechanism from spectra in collisions at TeV

Xing-rui Gou, Feng-lan Shao, Rui-qin Wang +2

We show that the experimental data of mid-rapidity spectra for proton, , , , and in minimum-bias collisions at $\…

cond-mat.mtrl-sci2025

Atomistic insight into the effects of solute and pressure on phase transformation in titanium alloys

Huicong Chen, Chenwei Shao, Zhuocheng Xie +2

The phase stability and transformation between hexagonal close-packed (hcp) α-phase and body-centered cubic (bcc) \b{eta}-phase in titanium (Ti) alloys are critical to their mecha…

cond-mat.mtrl-sci2017

Hydrogen bubble nucleation by self-clustering: Density Functional Theory and statistical models studies using tungsten as a model system

Jie Hou, Xiang-Shan Kong, Jingjing Sun +4

Low-energy hydrogen irradiation is known to induce bubble formation in tungsten, while its atomistic mechanisms remain little understood. Using first-principles calculations and st…

hep-ph2022

Production of single-charm hadrons by quark combination mechanism in -Pb collisions at TeV

Hai-hong Li, Feng-lan Shao, Jun Song +1

If QGP-like medium is created in -Pb collisions at extremely high collision energies, charm quarks that move in the medium can hadronize by capturing the co-moving light quark(s…

hep-ph2022

Collision centrality and energy dependence of strange hadron production in Au + Au collisions at \sqrt{s_{NN}}= 7.7-54.4 GeV

Yanting Feng, Ziyao Song, Fenglan Shao +1

We apply an equal-velocity quark combination model to systematically study the transverse momentum (p_{T}) spectra of strange hadrons K_{S}^{0}, ϕ, Λ, Ξ^{-}, Ω^{-}, \barΛ, \ba…

hep-ph2019

Constituent quark number scaling from strange hadron spectra in collisions at 13 TeV

Jian-wei Zhang, Hai-hong Li, Feng-lan Shao +1

We show that the data of spectra of and at midrapidity in inelastic events in collisions at 13 TeV exhibit a constituent quark number scaling…

cs.CV2025

Video SimpleQA: Towards Factuality Evaluation in Large Video Language Models

Meng Cao, Pengfei Hu, Yingyao Wang +12

Recent advancements in Large Video Language Models (LVLMs) have highlighted their potential for multi-modal understanding, yet evaluating their factual grounding in videos remains…

stat.ML2016

Parallel Chromatic MCMC with Spatial Partitioning

Jun Song, David A. Moore

We introduce a novel approach for parallelizing MCMC inference in models with spatially determined conditional independence relationships, for which existing techniques exploiting…

nucl-th2023

Production properties of deuterons, helions and tritons via an analytical nucleon coalescence method in Pb-Pb collisions at TeV

Rui-Qin Wang, Yan-Hao Li, Jun Song +1

We improve a nucleon coalescence model to include the coordinate-momentum correlation in nucleon joint distributions, and apply it to Pb-Pb collisions at TeV t…

physics.app-ph2024

Effect of particle oxidation, size and material on deformation, bonding and deposition during cold spray: a peridynamic investigation

Baihua Ren, Jun Song

Cold spray (CS) has emerged as an important additive manufacturing technology over the past decade. This study investigates the effect of oxide layers on the CS process, focusing o…

physics.optics2019

Photonic hooks from Janus microcylinders

Guoqiang Gu, Liyang Shao, Jun Song +8

Recently, a type of curved light beams, photonic hooks (PHs), was theoretically predicted and experimentally observed. The production of photonic hook (PH) is due to the breaking o…

cond-mat.mtrl-sci2020

Accurate prediction of nanovoid structures and energetics in bcc metals

Jie Hou, Yu-Wei You, Xiang-Shan Kong +2

Knowledge on structures and energetics of nanovoids is fundamental to understand defect evolution in metals. Yet there remain no reliable methods able to determine essential struct…

cs.CV2026

Global Compression Commander: Plug-and-Play Inference Acceleration for High-Resolution Large Vision-Language Models

Xuyang Liu, Ziming Wang, Junjie Chen +6

Large vision-language models (LVLMs) excel at visual understanding, but face efficiency challenges due to quadratic complexity in processing long multi-modal contexts. While token…

cond-mat.mtrl-sci2015

The Role of Low-lying Optical Phonons in Lattice Thermal Conductance of Rare-earth Pyrochlores: A First-principle Study

Guoqiang Lan, Bin Ouyang, Jun Song

Rare-earth pyrochlores, commonly exhibiting anomalously low lattice thermal conductivities, are considered as promising topcoat materials for thermal barrier coatings. However the…

hep-ph2023

Different Coalescence Sources of Light Nuclei Production in Au-Au Collisions at GeV

Rui-Qin Wang, Ji-Peng Lv, Yan-Hao Li +2

We study the production of light nuclei in the coalescence mechanism in Au-Au collisions at midrapidity at GeV. We derive analytic formulas of momentum distributi…

cs.CL2025

GeoSense: Evaluating Identification and Application of Geometric Principles in Multimodal Reasoning

Liangyu Xu, Yingxiu Zhao, Jingyun Wang +9

Geometry problem-solving (GPS), a challenging task requiring both visual comprehension and symbolic reasoning, effectively measures the reasoning capabilities of multimodal large l…

hep-ph2009

Centrality, system size and energy dependences of charged-particle pseudo-rapidity distribution

De-ming Wei, Feng-lan Shao, Jun Song +1

Utilizing the three-fireball picture within the quark combination model, we study systematically the charged particle pseudorapidity distributions in both Au+Au and Cu+Cu collision…

cond-mat.mtrl-sci2025

A Novel Discovery of Negative Thermal Expansion in Rare-earth Pyrochlore through Anion Order-Disorder Transition

Yuxuan Wang, Guoqiang Lan, Jun Song

In this study, we report for the first time the occurrence and investigation of the negative thermal expansion (NTE) effect in rare-earth pyrochlores. It is found that the NTE orig…

cs.CV2025

FILA: Fine-Grained Vision Language Models

Shiding Zhu, Wenhui Dong, Jun Song +3

Recently, there has been growing interest in the capability of multimodal large language models (MLLMs) to process high-resolution images. A common approach currently involves dyna…

cs.LG2020

Optimistic Distributionally Robust Policy Optimization

Jun Song, Chaoyue Zhao

Trust Region Policy Optimization (TRPO) and Proximal Policy Optimization (PPO), as the widely employed policy based reinforcement learning (RL) methods, are prone to converge to a…

hep-ph2010

Hadron yield correlation and constituent quark degree of freedom in heavy ion collisions

Rui-qin Wang, Feng-lan Shao, Jun Song +1

Based on the assumption of the production of deconfined quark matter, we use a quark combination model to systematically investigate hadron yields in heavy ion collisions from RHIC…

cs.LG2023

Towards Optimal Pricing of Demand Response -- A Nonparametric Constrained Policy Optimization Approach

Jun Song, Chaoyue Zhao

Demand response (DR) has been demonstrated to be an effective method for reducing peak load and mitigating uncertainties on both the supply and demand sides of the electricity mark…

cs.CV2026

Geoparsing: Diagram Parsing for Plane and Solid Geometry with a Unified Formal Language

Peijie Wang, Ming-Liang Zhang, Jun Cao +10

Multimodal Large Language Models (MLLMs) have achieved remarkable progress but continue to struggle with geometric reasoning, primarily due to the perception bottleneck regarding f…

cond-mat.mtrl-sci2024

Modulating dislocation reactions through preferential hydrogen segregation in bcc metals

Jie Hou, Ducheng Peng, Xiang-Shan Kong +4

The interaction between dislocations is fundamental to plastic deformation, work hardening, and defect accumulation. While extensive research has focused on the impact of solutes o…

cond-mat.mtrl-sci2015

Chemomechanical Origin of Hydrogen Trapping at Grain Boundaries in FCC Metals

Xiao Zhou, Daniel Marchand, David L. McDowell +2

Hydrogen embrittlement of metals is widely observed, but its atomistic origins remain little understood and much debated. Combining a unique identification of interstitial sites th…

cs.AI2026

How Foundational Skills Influence VLM-based Embodied Agents:A Native Perspective

Bo Peng, Pi Bu, Keyu Pan +7

Recent advances in vision-language models (VLMs) have shown promise for human-level embodied intelligence. However, existing benchmarks for VLM-driven embodied agents often rely on…

cs.CV2026

DDA-Thinker: Decoupled Dual-Atomic Reinforcement Learning for Reasoning-Driven Image Editing

Hanqing Yang, Qiang Zhou, Yongchao Du +6

Recent image editing models have achieved strong visual fidelity but often struggle with tasks requiring complex reasoning. To investigate and enhance the reasoning-grounded planni…

cond-mat.mtrl-sci2024

Understanding the Influence of Hydrogen on BCC Iron Grain Boundaries using the Kinetic Activation Relaxation technique (k-ART)

Aynour Khosravi, Jun Song, Normand Mousseau

Hydrogen embrittlement (HE) poses a significant challenge in the mechanical integrity of iron and its alloys. This study explores the influence of hydrogen atoms on two distinct gr…

hep-ph2015

Hadronization effects on the baryon-strangeness correlation in quark combination models

Feng-lan Shao, Jun Song, Rui-qin Wang

The baryon-strangeness correlation in the hadronization of the quark matter is studied within the quark combination mechanism. We calculate the correlation coefficient $C_{BS} = -3…

nucl-th2020

Quark number scaling of spectra for and in relativistic heavy-ion collisions

Jun Song, Feng-lan Shao, Zuo-tang Liang

We show that the experimental data of transverse momentum () spectra of baryon and meson at mid-rapidity in heavy-ion collisions exhibit the constituent quark numb…

nucl-th2022

Averaged transverse momentum correlations of hadrons in relativistic heavy-ion collisions

Yan-ting Feng, Feng-lan Shao, Jun Song

We compile experimental data for the averaged transverse momentum () of proton, , , and at mid-rapidity in Au+Au collisi…

cs.CV2025

Laser: Efficient Language-Guided Segmentation in Neural Radiance Fields

Xingyu Miao, Haoran Duan, Yang Bai +5

In this work, we propose a method that leverages CLIP feature distillation, achieving efficient 3D segmentation through language guidance. Unlike previous methods that rely on mult…

cs.AI2024

Can VLMs Play Action Role-Playing Games? Take Black Myth Wukong as a Study Case

Peng Chen, Pi Bu, Jun Song +2

Recently, large language model (LLM)-based agents have made significant advances across various fields. One of the most popular research areas involves applying these agents to vid…

cs.LG2024

Enhancing the Resilience of Graph Neural Networks to Topological Perturbations in Sparse Graphs

Shuqi He, Jun Zhuang, Ding Wang +2

Graph neural networks (GNNs) have been extensively employed in node classification. Nevertheless, recent studies indicate that GNNs are vulnerable to topological perturbations, suc…

cs.CV2024

Dreamer XL: Towards High-Resolution Text-to-3D Generation via Trajectory Score Matching

Xingyu Miao, Haoran Duan, Varun Ojha +4

In this work, we propose a novel Trajectory Score Matching (TSM) method that aims to solve the pseudo ground truth inconsistency problem caused by the accumulated error in Interval…

cs.AI2026

Crisis-Bench: Benchmarking Strategic Ambiguity and Reputation Management in Large Language Models

Cooper Lin, Maohao Ran, Yanting Zhang +6

Standard safety alignment optimizes Large Language Models (LLMs) for universal helpfulness and honesty, effectively instilling a rigid "Boy Scout" morality. While robust for genera…

cs.AI2026

CaveAgent: Transforming LLMs into Stateful Runtime Operators

Maohao Ran, Zhenglin Wan, Cooper Lin +21

LLM-based agents are increasingly capable of complex task execution, yet current agentic systems remain constrained by text-centric paradigms that struggle with long-horizon tasks…

cond-mat.mtrl-sci2024

First Principles Studies of Stacking Fault Energies in Ternary Magnesium Alloys

Qiwen Qiu, Stephen Yue, Jun Song

Magnesium (Mg) alloys have emerged as promising materials due to their low density and high strength-to-weight ratio, offering a wide range of applications across multiple industri…

hep-ph2017

Different production sources of light nuclei in relativistic heavy-ion collisions

Rui-qin Wang, Jun Song, Gang Li +1

We propose a new method, i.e., an exclusive quark combination model + an inclusive hadron recombination model, to study different production sources of light nuclei in relativistic…

hep-ph2017

Yield ratios of identified hadrons in p+p, p+Pb, Pb+Pb collisions at the Large Hadron Collider

Feng-lan Shao, Guo-jing Wang, Rui-qin Wang +2

Yield ratios of identified hadrons observed in high multiplicity p+p and p+Pb collisions at LHC show remarkable similarity with those in Pb+Pb collisions, indicating some important…

cs.AI2026

MOSS: Self-Evolution through Source-Level Rewriting in Autonomous Agent Systems

Qianshu Cai, Yonggang Zhang, Xianzhang Jia +5

Autonomous agentic systems are largely static after deployment: they do not learn from user interactions, and recurring failures persist until the next human-driven update ships a…

cond-mat.mtrl-sci2024

Molecular dynamics simulations of interaction between a super edge dislocation and interstitial dislocation loops in irradiated L12-Ni3Al

Cheng Chen, Dongyang Qin, Yiding Wang +2

The study employed MD simulations to investigate the interactions between a <110> super-edge dislocation, consisting of the four Shockley partials, and interstitial dislocation loo…

cs.CV2026

AdaGen: Learning Adaptive Policy for Image Synthesis

Zanlin Ni, Yulin Wang, Yeguo Hua +5

Recent advances in image synthesis have been propelled by powerful generative models, such as Masked Generative Transformers (MaskGIT), autoregressive models, diffusion models, and…

cs.IR2024

Fine-Grained Embedding Dimension Optimization During Training for Recommender Systems

Qinyi Luo, Penghan Wang, Wei Zhang +8

Huge embedding tables in modern deep learning recommender models (DLRM) require prohibitively large memory during training and inference. This paper proposes FIITED, a system to au…

cs.CL2026

Scaling Multi-Hop Training Data via Graph-Constrained Path Selection

Pengyu Chen, Yonggang Zhang, Mingming Chen +3

Endowing large language models with compositional reasoning over specialized documents requires multi-hop training data at scale, where such data rarely exists outside of curated b…

math.OC2023

Distributionally Robust Optimal Power Flow with Uncertain Renewable Energy Output

Jia Yang, Jun Song, Chaoyue Zhao

Optimal power flow (OPF) is an important tool for Independent System Operators (ISOs) to deal with the power generation management. With the increasing penetration of renewable ene…

hep-ph2009

Centrality dependence of spectra for identified hadrons in Au+Au and Cu+Cu collisions at GeV

Yun-fei Wang, Feng-lan Shao, Jun Song +2

The centrality dependence of transverse momentum spectra for identified hadrons at midrapidity in Au+Au collisions at GeV is systematically studied in a quark…

cs.AI2026

Mixture-of-Visual-Thoughts: Exploring Context-Adaptive Reasoning Mode Selection for General Visual Reasoning

Zejun Li, Yingxiu Zhao, Jiwen Zhang +6

Current visual reasoning methods mainly focus on exploring specific reasoning modes. Although improvements can be achieved in particular domains, they struggle to develop general r…

cs.AI2024

Align Anything: Training All-Modality Models to Follow Instructions with Language Feedback

Jiaming Ji, Jiayi Zhou, Hantao Lou +16

Reinforcement learning from human feedback (RLHF) has proven effective in enhancing the instruction-following capabilities of large language models; however, it remains underexplor…

nucl-th2019

Statistical method in quark combination model

Yang-guang Yang, Jun Song, Feng-lan Shao +2

We present a new method of solving the probability distribution for baryons, antibaryons and mesons at the hadronization of constituent quark and antiquark system. The hadronizatio…

hep-ph2009

Hadron production by quark combination in central Pb+Pb collisions at GeV

Chang-en Shao, Jun Song, Feng-lan Shao +1

The quark combination mechanism of QGP hadronization is applied to nucleus-nucleus collisions at top SPS energy. The yields, rapidity and transverse momentum distributions of ident…

hep-ph2018

New feature of low charm quark hadronization in collisions at TeV

Jun Song, Hai-hong Li, Feng-lan Shao

Treating the light-flavor constituent quarks and antiquarks that can well describe the data of light-flavor hadrons in collisions at TeV as the underlying source…

cond-mat.mtrl-sci2024

Critical Assessment of Mass and Lattice Disorder in Thermal Conductivity Prediction for Medium and High Entropy Ceramics

Yuxuan Wang, Guoqiang Lan, Jun Song

Medium and high entropy ceramics, with their distinctive disordered structures, exhibit ultra-low thermal conductivity and high temperature stability. These properties make them st…

hep-ph2017

Quark number scaling of hadronic spectra and constituent quark degree of freedom in -Pb collisions at TeV

Jun Song, Xing-rui Gou, Feng-lan Shao +1

We show that the experimental data of spectra of identified hadrons released recently by ALICE collaboration for -Pb collisions at TeV exhibit a disti…

nucl-th2017

Multiplicity fluctuation and correlation of identified baryons in quark combination model

Jun Song, Hai-hong Li, Rui-qin Wang +1

The dynamical fluctuation and correlation of multiplicity distributions of identified baryons and antibaryons produced by the hadronization of the bulk quark system are systematica…

nucl-th2019

Rapidity dependence of global polarization in heavy ion collisions

Zuo-Tang Liang, Jun Song, Isaac Upsal +2

We use a geometric model for the hadron polarization with an emphasis on the rapidity dependence. It is based on the model of Brodsky, Gunion, and Kuhn and that of the Bjorken scal…

hep-ph2026

Anisotropic flows of light-flavor and charmed hadrons in Pb+Pb collisions at LHC energy

Yan-ting Feng, Rui-qin Wang, Feng-lan Shao +1

We apply a constituent quark equal-velocity combination (EVC) model to study the elliptic flow () and triangular flow () of light-flavor and single-charmed hadrons in…

cond-mat.mtrl-sci2020

A novel approach to assess hydrogen embrittlement (HE) susceptibility and mechanisms in high strength martensitic steels

Tuhin Das, Salim V. Brahimi, Jun Song +1

A rapid fracture test in four-point bending is proposed to assess hydrogen embrittlement (HE) susceptibility of high strength martensitic steels. The novelty of this technique is t…

nucl-th2017

Multiplicity fluctuation and correlation of mesons and baryons in ultra-relativistic heavy-ion collisions at LHC

Hai-hong Li, Feng-lan Shao, Jun Song

We study the multiplicity fluctuation and correlation of identified mesons and baryons formed at the hadronization by the quark combination mechanism in the context of ultra-relati…

cs.CV2025

MMG-Vid: Maximizing Marginal Gains at Segment-level and Token-level for Efficient Video LLMs

Junpeng Ma, Qizhe Zhang, Ming Lu +4

Video Large Language Models (VLLMs) excel in video understanding, but their excessive visual tokens pose a significant computational challenge for real-world applications. Current…

cs.CV2026

Thinking with Drafts: Speculative Temporal Reasoning for Efficient Long Video Understanding

Pengfei Hu, Meng Cao, Yingyao Wang +6

Long video understanding is essential for human-like intelligence, enabling coherent perception and reasoning over extended temporal contexts. While the emerging thinking-with-fram…

hep-ph2026

Transverse momentum dependence of ratio in high energy collisions

Hai-hong Li, Jun Song, Feng-lan Shao

We apply a constituent quark equal-velocity combination model to study the dependence of ratio in , -Pb and Pb-Pb collisions at LHC energies. We demonstrate…

cs.AI2026

InquireMobile: Teaching VLM-based Mobile Agent to Request Human Assistance via Reinforcement Fine-Tuning

Qihang Ai, Pi Bu, Yue Cao +8

Recent advances in Vision-Language Models (VLMs) have enabled mobile agents to perceive and interact with real-world mobile environments based on human instructions. However, the c…

cs.CV2026

SecAgent: Efficient Mobile GUI Agent with Semantic Context

Yiping Xie, Song Chen, Jingxuan Xing +7

Mobile Graphical User Interface (GUI) agents powered by multimodal large language models have demonstrated promising capabilities in automating complex smartphone tasks. However, e…

cs.AI2025

AI Deception: Risks, Dynamics, and Controls

Boyuan Chen, Sitong Fang, Jiaming Ji +56

As intelligence increases, so does its shadow. AI deception, in which systems induce false beliefs to secure self-beneficial outcomes, has evolved from a speculative concern to an…

cs.CV2025

FiLA-Video: Spatio-Temporal Compression for Fine-Grained Long Video Understanding

Yanan Guo, Wenhui Dong, Jun Song +7

Recent advancements in video understanding within visual large language models (VLLMs) have led to notable progress. However, the complexity of video data and contextual processing…

nucl-th2021

Elliptic flow of hadrons in equal-velocity quark combination mechanism in relativistic heavy-ion collisions

Jun Song, Hai-hong Li, Feng-lan Shao

We apply a quark combination model with equal-velocity combination (EVC) approximation to study the elliptic flow () of hadrons in heavy-ion collisions in a wide collision e…

cs.CV2026

ViT: Unlocking Test-Time Training in Vision

Dongchen Han, Yining Li, Tianyu Li +6

Test-Time Training (TTT) has recently emerged as a promising direction for efficient sequence modeling. TTT reformulates attention operation as an online learning problem, construc…

cs.CV2026

MOON: Generative MLLM-based Multimodal Representation Learning for E-commerce Product Understanding

Daoze Zhang, Chenghan Fu, Zhanheng Nie +7

With the rapid advancement of e-commerce, exploring general representations rather than task-specific ones has attracted increasing research attention. For product understanding, a…

hep-ph2012

Hadron Yield Correlation in Combination Models in High Energy AA Collisions

Rui-qin Wang, Feng-lan Shao, Jun Song +2

We study the hadron yield correlation in the combination models in high energy heavy ion collisions. We derive the relationship between the average yields of different hadrons prod…

cs.CV2026

Deep Pre-Alignment for VLMs

Tianyu Yu, Kechen Fang, Zihao Wan +5

Most Vision Language Models (VLMs) directly map outputs from ViT encoders to the LLM via a lightweight projector. While effective, recent analysis suggests this architecture suffer…

cs.LG2023

Provably Convergent Policy Optimization via Metric-aware Trust Region Methods

Jun Song, Niao He, Lijun Ding +1

Trust-region methods based on Kullback-Leibler divergence are pervasively used to stabilize policy optimization in reinforcement learning. In this paper, we exploit more flexible m…

cond-mat.mtrl-sci2015

Energetics and kinetics of vacancies in monolayer graphene boron nitride heterostructures

Bin Ouyang, Fanchao Meng, Jun Song

Graphene and boron nitride (GPBN) heterostructures provide a viable way to realize tunable bandgap, promising new opportunities in graphene-based nanoelectronic and optoelectronic…

cond-mat.mtrl-sci2024

Kinetics of hydrogen and vacancy diffusion in iron: A Kinetic Activation Relaxation technique (k-ART) study

Aynour Khosravi, Jun Song, Normand Mousseau

We investigate hydrogen (H) and mono and divacancy-hydrogen complexes (VH and VH) diffusion in body-centered-cubic (BCC) iron using the kinetic Activation-Relaxation Te…

hep-ph2021

Production of light flavor and single-charmed hadrons in collisions at TeV in an equal-velocity quark combination model

Hai-hong Li, Feng-lan Shao, Jun Song

We apply an equal-velocity quark combination model to study the production of light-flavor hadrons and single-charmed hadrons at midrapidity in collisions at T…

hep-ph2019

Charmed hadron production via equal-velocity quark combination in ultra-relativistic heavy ion collisions

Rui-Qin Wang, Jun Song, Feng-Lan Shao +1

Recent data on the production of mesons and baryons in heavy ion collisions at the Relativistic Heavy Ion Collider and the Large Hadron Collider exhibit a number of st…

hep-ph2013

Yield correlations and dependence of charm hadrons in Pb + Pb collisions at TeV

Rui-qin Wang, Jun Song, Feng-lan Shao

Recently the ALICE Collaboration has published a few data of charm hadrons in Pb + Pb collisions at the CERN Large Hadron Collider (LHC). We extend the quark combination to the cha…

cs.CL2026

The Flexibility Trap: Rethinking the Value of Arbitrary Order in Diffusion Language Models

Zanlin Ni, Shenzhi Wang, Yang Yue +8

Diffusion Large Language Models (dLLMs) break the rigid left-to-right constraint of traditional LLMs, enabling token generation in arbitrary orders. Intuitively, this flexibility i…

stat.ML2024

Enhancing Sufficient Dimension Reduction via Hellinger Correlation

Seungbeom Hong, Ilmun Kim, Jun Song

In this work, we develop a new theory and method for sufficient dimension reduction (SDR) in single-index models, where SDR is a sub-field of supervised dimension reduction based o…

cs.CV2024

GeoGPT4V: Towards Geometric Multi-modal Large Language Models with Geometric Image Generation

Shihao Cai, Keqin Bao, Hangyu Guo +3

Large language models have seen widespread adoption in math problem-solving. However, in geometry problems that usually require visual aids for better understanding, even the most…

cs.CV2026

SPAE: Spectrally Guided Autoencoder for Pretrained Visual Latents

Yibin Huang, Jixiang Hong, Zongzhao Li +8

Latents from vision foundation models (VFMs) are semantically rich and well suited for visual understanding. Recent representation autoencoder methods such as RAE have shown that t…

nucl-th2021

Strange hadron production in a quark combination model in Au+Au collisions at RHIC energies

Jun Song, Xiao-feng Wang, Hai-hong Li +2

We apply a quark combination model to study yield densities and transverse momentum () spectra of strange (anti-)hadrons at mid-rapidity in central Au+Au collisions at $\sqr…

hep-ph2016

Hidden-charm Pentaquark States in Heavy Ion Collisions at the Large Hadron Collider

Rui-Qin Wang, Jun Song, Kai-Jia Sun +3

In the framework of the quark combination, we derive the yield formulas and study the yield ratios of the hidden-charm pentaquark states in ultra-relativistic heavy ion collisions.…

cs.CV2026

CombatVLA: An Efficient Vision-Language-Action Model for Combat Tasks in 3D Action Role-Playing Games

Peng Chen, Pi Bu, Yingyao Wang +9

Recent advances in Vision-Language-Action models (VLAs) have expanded the capabilities of embodied intelligence. However, significant challenges remain in real-time decision-making…

cs.CV2024

Demystify Mamba in Vision: A Linear Attention Perspective

Dongchen Han, Ziyi Wang, Zhuofan Xia +7

Mamba is an effective state space model with linear computation complexity. It has recently shown impressive efficiency in dealing with high-resolution inputs across various vision…

cs.AI2026

Pailitao-MMSearch: Building Native E-Commerce Multimodal Search Foundation

Xiaohan Ye, Xu Chen, Zihan Gong +15

The evolution of e-commerce has fundamentally transformed how users search for products, shifting from simple text-based keyword queries to complex multimodal interactions that sea…

cond-mat.mtrl-sci2020

Hydrogen clustering in bcc metals: atomic origin and strong stress anisotropy

Jie Hou, Xiang-Shan Kong, C. S. Liu +1

Hydrogen (H) induced damage in metals has been a long-standing woe for many industrial applications. One form of such damage is linked to H clustering, for which the atomic origin…

cs.CV2026

GIFT: Global Irreplaceability Frame Targeting for Efficient Video Understanding

Junpeng Ma, Sashuai Zhou, Guanghao Li +9

Video Large Language Models (VLMs) have achieved remarkable success in video understanding, but the significant computational cost from processing dense frames severely limits thei…

cs.AI2026

Text Before Vision: Staged Knowledge Injection Matters for Agentic RLVR in Ultra-High-Resolution Remote Sensing Understanding

Fengxiang Wang, Mingshuo Chen, Yueying Li +13

Multimodal reasoning for ultra-high-resolution (UHR) remote sensing (RS) is usually bottlenecked by visual evidence acquisition: the model necessitates localizing tiny task-relevan…

hep-ph2011

Hadronic Rapidity Spectra in Heavy Ion Collisions at SPS and AGS energies in a Quark Combination Model

Le-xue Sun, Rui-qin Wang, Jun Song +1

The quark combination mechanism of hadron production is applied to nucleus-nucleus collisions at the CERN Super Proton Synchrotron (SPS) and BNL Alternating Gradient Synchrotron (A…

stat.ME2021

Multivariate functional group sparse regression: functional predictor selection

Ali Mahzarnia, Jun Song

In this paper, we propose methods for functional predictor selection and the estimation of smooth functional coefficients simultaneously in a scalar-on-function regression problem…

cs.AI2026

Mobile-R1: Towards Interactive Capability for VLM-Based Mobile Agent via Systematic Training

Jihao Gu, Qihang Ai, Yingyao Wang +10

Vision-language model-based mobile agents have gained the ability to understand complex instructions and mobile screenshots, benefiting from reinforcement learning paradigms like G…

cs.CV2026

MatSegNet: a New Boundary-aware Deep Learning Model for Accurate Carbide Precipitate Analysis in High-Strength Steels

Xiaohan Bie, Manoj Arthanari, Evelin Barbosa de Melo +7

Lower Bainite (LB) and Tempered Martensite (TM) are two common microstructures in modern high-strength steels. LB and TM can render similar mechanical properties for steels, yet LB…

cs.AI2026

Reinforcement Learning of Large Language Models for Interpretable Credit Card Fraud Detection

Cooper Lin, Yanting Zhang, Maohao Ran +7

E-commerce platforms and payment solution providers face increasingly sophisticated fraud schemes, ranging from identity theft and account takeovers to complex money laundering ope…

cs.SD2025

Masked Self-distilled Transducer-based Keyword Spotting with Semi-autoregressive Decoding

Yu Xi, Xiaoyu Gu, Haoyu Li +3

RNN-T-based keyword spotting (KWS) with autoregressive decoding~(AR) has gained attention due to its streaming architecture and superior performance. However, the simplicity of the…

cond-mat.mtrl-sci2015

Covalent pathways in engineering h-BN supported graphene

Bin Ouyang, Jun Song

Cross-planar di-vacancies (CPDVs) within stacked graphene hexagonal boron nitride (h-BN) heterostructures provide stabilized covalent links to bridge adjacent graphene and h-BN she…

hep-ph2025

The domain of soft hadrons in transverse momentum space in high energy collisions

Jun Song, Hai-hong Li, Feng-lan Shao

By studying experimental data for invariant transverse momentum distribution of hadrons in high energy , A and AA collisions, we find two characteristic p…

hep-ph2024

Production characteristics of light nuclei, hypertritons and -hypernuclei in Pb+Pb collisions at TeV

Rui-Qin Wang, Xin-Lei Hou, Yan-Hao Li +2

We extend an analytical nucleon coalescence model with hyperons to study productions of light nuclei, hypertritons and -hypernuclei in Pb+Pb collisions at