Publications (73)
AVID: A Benchmark for Omni-Modal Audio-Visual Inconsistency Understanding via Agent-Driven Construction
Zixuan Chen, Depeng Wang, Hao Lin +6
We present AVID, the first large-scale benchmark for audio-visual inconsistency understanding in videos. While omni-modal large language models excel at temporally aligned tasks su…
Properties of the limit shape for some last passage growth models in random environments (Dissertation)
Hao Lin
We study directed last-passage percolation on the planar square lattice whose weights have general distributions, or equivalently, queues in series with general service distributio…
Sharp Diagonal Thresholds for Tight Hamilton Cycles in Uniformly Dense -Graphs
Hao Lin, Guanghui Wang, Wenling Zhou
A -uniform hypergraph (or -graph) on vertices is \emph{-dense} if for all . This is one of the weakest…
Dual Stream Computer-Generated Image Detection Network Based On Channel Joint And Softpool
Ziyi Xi, Hao Lin, Weiqi Luo
With the development of computer graphics technology, the images synthesized by computer software become more and more closer to the photographs. While computer graphics technology…
TIT: A Tree-Structured Instruction Tuning Approach for LLM-Based Code Translation
He Jiang, Yufu Wang, Hao Lin +5
Large Language Models (LLMs) have shown strong performance in automated source-to-target code translation through pretraining on extensive code corpora. However, mainstream LLM-bas…
A strengthening on consecutive odd cycles in graphs of given minimum degree
Hao Lin, Guanghui Wang, Wenling Zhou
Liu and Ma [J. Combin. Theory Ser. B, 2018] conjectured that every -connected non-bipartite graph with minimum degree at least contains cycles with con…
Bridging Traffic State and Trajectory for Dynamic Road Network and Trajectory Representation Learning
Chengkai Han, Jingyuan Wang, Yongyao Wang +4
Effective urban traffic management is vital for sustainable city development, relying on intelligent systems with machine learning tasks such as traffic flow prediction and travel…
Connectivity of LEO Satellite Mega Constellations: An Application of Percolation Theory on a Sphere
Hao Lin, Mustafa A. Kishk, Mohamed-Slim Alouini
With the advent of the 6G era, global connectivity has become a common goal in the evolution of communications, aiming to bring Internet services to more unconnected regions. Addit…
Quantification of propidium iodide delivery with millisecond electric pulses: A model study
Miao Yu, Hao Lin
A model study of propidium iodide delivery with millisecond electric pulses is presented; this work is a companion of the experimental efforts by Sadik et al. [1]. Both membrane pe…
Extremal problems in uniformly dense hypergraphs and digraphs
Hao Lin, Guanghui Wang, Wenling Zhou +1
The uniform Turán density of a -uniform hypergraph (or -graph) is the supremum of all such that there exist infinitely many -free -graphs in wh…
Enhancing LLM-Based Code Translation with Verified Multi-Semantic Representations
Yufu Wang, He Jiang, Hao Lin +4
Large language models (LLMs) have shown great promise for automated code translation, yet existing approaches often rely on token-level statistical patterns rather than sufficient…
DriveAgent: Multi-Agent Structured Reasoning with LLM and Multimodal Sensor Fusion for Autonomous Driving
Xinmeng Hou, Wuqi Wang, Long Yang +4
We introduce DriveAgent, a novel multi-agent autonomous driving framework that leverages large language model (LLM) reasoning combined with multimodal sensor fusion to enhance situ…
Unraveling Intertwined Orders in the Strongly Correlated Kagome Metal CsCr3Sb5
Liangyang Liu, Yidian Li, Hengxin Tan +11
While correlated phenomena of flat bands have been extensively studied in twisted systems, the ordered states that emerge from interactions in the intrinsic flat bands of kagome la…
Share More, Search Less: Collaborative Parallel Thinking for Efficient Test-Time Scaling
Xinglin Wang, Hao Lin, Shaoxiong Feng +9
Test-Time Scaling (TTS) enhances the reasoning capabilities of large language models by allocating additional inference compute to explore the solution space. However, existing par…
Energy-as-a-Service for RF-Powered IoE Networks: A Percolation Theory Approach
Hao Lin, Ainur Zhaikhan, Mustafa A. Kishk +2
Due to the involved massive number of devices, radio frequency (RF) energy harvesting is indispensable to realize the foreseen Internet-of-Everything (IoE) within 6G networks. Anal…
Navigating the Risks of Using Large Language Models for Text Annotation in Social Science Research
Hao Lin, Yongjun Zhang
Large language models (LLMs) have the potential to revolutionize computational social science, particularly in automated textual analysis. In this paper, we conduct a systematic ev…
A transient solution for droplet deformation under electric fields
Jia Zhang, Jeffery D. Zahn, Hao Lin
A transient analysis to quantify droplet deformation under DC electric fields is presented. The full Taylor-Melcher leaky dielectric model is employed where the charge relaxation t…
Accurate Indoor Localization Using Acoustic Direction Finding via Smart Phones
Wenchao Huang, Yan Xiong, Xiang-Yang Li +4
We propose and implement a novel indoor localization scheme, Swadloon, built upon an accurate acoustic direction finding. Swadloon leverages sensors of the smartphone without the r…
SurfAAV: Design and Implementation of a Novel Multimodal Surfing Aquatic-Aerial Vehicle
Kun Liu, Junhao Xiao, Hao Lin +4
Despite significant advancements in the research of aquatic-aerial robots, existing configurations struggle to efficiently perform underwater, surface, and aerial movement simultan…
Turán density of stars in uniformly dense hypergraphs
Hao Lin, Wenling Zhou
A -uniform hypergraph (or -graph) is -\emph{dense} if for any subsets , the number of triples such that $…
Decoupling anomaly discrimination and representation learning: self-supervised learning for anomaly detection on attributed graph
YanMing Hu, Chuan Chen, BoWen Deng +4
Anomaly detection on attributed graphs is a crucial topic for its practical application. Existing methods suffer from semantic mixture and imbalance issue because they mainly focus…
RollingQ: Reviving the Cooperation Dynamics in Multimodal Transformer
Haotian Ni, Yake Wei, Hang Liu +4
Multimodal learning faces challenges in effectively fusing information from diverse modalities, especially when modality quality varies across samples. Dynamic fusion strategies, s…
Collaborative Company Profiling: Insights from an Employee's Perspective
Hao Lin, Hengshu Zhu, Yuan Zuo +3
Company profiling is an analytical process to build an indepth understanding of company's fundamental characteristics. It serves as an effective way to gain vital information of th…
Hypergraphs with a quarter uniform Turán density
Hao Li, Hao Lin, Guanghui Wang +1
The uniform Turán density of a -uniform hypergraph is the supremum over all for which there is an -free hypergraph with the property that every linearly…
Knowing but Not Correcting: Routine Task Requests Suppress Factual Correction in LLMs
Zixuan Chen, Hao Lin, Zizhe Chen +6
LLMs reliably correct false claims when presented in isolation, yet when the same claims are embedded in task-oriented requests, they often comply rather than correct. We term this…
UMAIR-FPS: User-aware Multi-modal Animation Illustration Recommendation Fusion with Painting Style
Yan Kang, Hao Lin, Mingjian Yang +1
The rapid advancement of high-quality image generation models based on AI has generated a deluge of anime illustrations. Recommending illustrations to users within massive data has…
Domino: Decoupling Causal Modeling from Autoregressive Drafting in Speculative Decoding
Jianuo Huang, Yaojie Zhang, Qituan Zhang +3
Speculative decoding accelerates LLM inference by drafting multiple tokens and verifying them in parallel with the target model. However, its practical speedup is constrained by th…
Terahertz frequency upconversion by coherently driving charge dynamics in the InSb/CdTe heterostructure
Pai Peng, Mingxiang Pan, Jiuming Liu +10
We investigate terahertz (THz) harmonic generation in the InSb/CdTe heterostructure, demonstrating, for the first time, efficient in-plane magnetic field-induced second-harmonic ge…
E2 distribution and statistical regularity in polygonal planar tessellations
Ran Li, Consuelo Ibar, Zhenru Zhou +4
From solar supergranulation to salt flat in Bolivia, from veins on leaves to cells on Drosophila wing discs, polygon-based networks exhibit great complexities, yet similarities per…
Topological one-way Weyl fiber
Hao Lin, Yu Wang, Zitao Ji +3
Topological photonics enables unprecedented photon manipulation by realizing various topological states, such as corner states, edge states, and surface states. However, achieving…
No Query, No Access
Wenqiang Wang, Siyuan Liang, Yangshijie Zhang +3
Textual adversarial attacks mislead NLP models, including Large Language Models (LLMs), by subtly modifying text. While effective, existing attacks often require knowledge of the v…
Towards the Standardization of Non-orthogonal Multiple Access for Next Generation Wireless Networks
Yan Chen, Alireza Bayesteh, Yiqun Wu +13
Non-orthogonal multiple access (NoMA) as an efficient way of radio resource sharing can root back to the network information theory. For generations of wireless communication syste…
Performance Evaluation of RF-powered IoT in Rural Areas: The Wireless Power Digital Divide
Hao Lin, Mustafa A. Kishk, Mohamed-Slim Alouini
Bridging the digital divide is one of the goals of mobile networks in the future, and further building IoT networks in rural areas is a feasible solution. This paper studies the do…
Performance Analysis of Infrastructure Sharing Techniques in Cellular Networks: A Percolation Theory Approach
Hao Lin, Mustafa A. Kishk, Mohamed-Slim Alouini
In the context of 5G, infrastructure sharing has been identified as a potential solution to reduce the investment costs of cellular networks. In particular, it can help low-income…
Spinodal Instability at the Onset of Collective Expansion in Nuclear Collisions
Pawel Danielewicz, Hao Lin, Jirina R. Stone +1
Using transport theory to model central Au + Au collisions in the energy region of 20 - 110 MeV/u, at impact parameters b <= 5 fm, we predict a measurable impact of spinoidal insta…
Properties of the limit shape for some last passage growth models in random environments
Hao Lin, Timo Seppäläinen
We study directed last passage percolation on the first quadrant of the planar square lattice whose weights have general distributions, or equivalently, ./G/1 queues in series. The…
Earthquake-Like Avalanches in Compression of Ductile Porous Materials
Hao Lin, HaiYing Wang, Chunsheng Luc +1
Earthquakes are complex phenomena characterized by some fundamental seismic laws. However, under the framework of brittle fracture in disordered material, these universal scaling b…
Multi-task Adversarial Attacks against Black-box Model with Few-shot Queries
Wenqiang Wang, Yan Xiao, Hao Lin +2
Current multi-task adversarial text attacks rely on abundant access to shared internal features and numerous queries, often limited to a single task type. As a result, these attack…
Discovery of terahertz-frequency orbitally-coupled magnons in a kagome ferromagnet
Mengqian Che, Weizhao Chen, Maoyuan Wang +12
In ferromagnetic materials, magnons - quanta of spin waves - typically resonate in the gigahertz range. Beyond conventional magnons, while theoretical studies have predicted magnon…
Dynamics of one-dimensional correlated nuclear systems within non-equilibrium Green's function theory
Hao Lin, Hossein Mahzoon, Arnau Rios +1
Theory of non-equilibrium Green's function (NGF) provides a practical framework for studying quantum many-body systems out of equilibrium. Extending the previous mean field approac…
HAPS-enabled Downlink Coverage Enhancement in Islands and Maritime Areas
Hao Lin, Mustafa A. Kishk, Mohamed-Slim Alouini
Non-terrestrial networks (NTNs) are poised to play a critical role in next-generation mobile communications, offering enhanced flexibility, improved line-of-sight (LoS) conditions,…
FUSCO: High-Performance Distributed Data Shuffling via Transformation-Communication Fusion
Zhuoran Zhu, Chunyang Zhu, Hao Lin +9
Large-scale Mixture-of-Experts (MoE) models rely on \emph{expert parallelism} for efficient training and inference, which splits experts across devices and necessitates distributed…
Dirac-vortex topological photonic crystal fibre
Hao Lin, Ling Lu
The success of photonic crystal fibres relies largely on the endless variety of two-dimensional photonic crystals in the cross-section. Here, we propose a topological bandgap fibre…
MEGA: Second-Order Gradient Alignment for Catastrophic Forgetting Mitigation in GFSCIL
Jinhui Pang, Changqing Lin, Hao Lin +4
Graph Few-Shot Class-Incremental Learning (GFSCIL) enables models to continually learn from limited samples of novel tasks after initial training on a large base dataset. Existing…
RLinf-VLA: A Unified and Efficient Framework for Reinforcement Learning of Vision-Language-Action Models
Hongzhi Zang, Mingjie Wei, Si Xu +15
Recent studies have demonstrated the potential of reinforcement learning (RL) to improve the task performance of vision-language-action (VLA) models through interaction. However, c…
Non-Grotthuss Proton Diffusion Mechanism in Tungsten Oxide Dihydrate from First-Principles Calculations
Hao Lin, Fei Zhou, Chi-Ping Liu +1
Fast proton conduction mechanism is of key importance for achieving high performance in fuel cell membranes, batteries, supercapacitors, and electrochromic materials. Enhanced prot…
RLinf-USER: A Unified and Extensible System for Real-World Online Policy Learning in Embodied AI
Hongzhi Zang, Shu'ang Yu, Hao Lin +14
Online policy learning directly in the physical world is a promising yet challenging direction for embodied intelligence. Unlike simulation, real-world systems cannot be arbitraril…
Three matching intersection property for matching covered graphs
Hao Lin, Xiumei Wang
In connection with Fulkerson's conjecture on cycle covers, Fan and Raspaud proposed a weaker conjecture: For every bridgeless cubic graph , there are three perfect matchings $M_…
Realization of a Dirac-vortex topological photonic crystal fiber
Quanhao Niu, Bei Yan, Lei Shen +10
Photonic crystal fibers (PCFs) that trap and guide light using photonic bandgaps have revolutionized modern optics with enormous scientific innovations and technological applicatio…
One-body Langevin dynamics in heavy-ion collisions at intermediate energies
Hao Lin, Pawel Danielewicz
We present a new framework to treat the dissipation and fluctuation dynamics associated with nucleon-nucleon scattering in heavy-ion collisions. The two-body collision processes ar…
Nonlinear dynamic Process of Fluvial Process Based on Theories of Chaos and Dissipative Structure
Hao Lin
The analysis of riverbed evolution is an important basis for decision-making on river improvement, flood control project construction, and water conservancy hub dispatching and app…
STAlloc: Enhancing Memory Efficiency in Large-Scale Model Training with Spatio-Temporal Planning
Zixiao Huang, Junhao Hu, Hao Lin +9
The rapid scaling of large language models (LLMs) has significantly increased GPU memory pressure, which is further aggravated by training optimization techniques such as virtual p…
RLinf: Flexible and Efficient Large-scale Reinforcement Learning via Macro-to-Micro Flow Transformation
Chao Yu, Yuanqing Wang, Zhen Guo +26
Reinforcement learning (RL) has demonstrated immense potential in advancing artificial general intelligence, agentic intelligence, and embodied intelligence. However, the inherent…
Not Just What's There: Enabling CLIP to Comprehend Negated Visual Descriptions Without Fine-tuning
Junhao Xiao, Zhiyu Wu, Hao Lin +5
Vision-Language Models (VLMs) like CLIP struggle to understand negation, often embedding affirmatives and negatives similarly (e.g., matching "no dog" with dog images). Existing me…
VISD: Enhancing Video Reasoning via Structured Self-Distillation
Hao Lin, Kunyang Lv, Xu Jiang +5
Training VideoLLMs for complex reasoning remains challenging due to sparse sequence level rewards and the lack of fine grained credit assignment over long, temporally grounded reas…
: Online RL Fine-tuning for Flow-based Vision-Language-Action Models
Kang Chen, Zhihao Liu, Tonghe Zhang +11
Vision-Language-Action (VLA) models enable robots to understand and perform complex tasks from multimodal input. Although recent work explores using reinforcement learning (RL) to…
Improved Xception with Dual Attention Mechanism and Feature Fusion for Face Forgery Detection
Hao Lin, Weiqi Luo, Kangkang Wei +1
With the rapid development of deep learning technology, more and more face forgeries by deepfake are widely spread on social media, causing serious social concern. Face forgery det…
A transient solution for vesicle electrodeformation and relaxation
Jia Zhang, Jeffery D. Zahn, Wenchang Tan +1
A transient analysis for vesicle deformation under DC electric fields is developed. The theory extends from a droplet model, with the additional consideration of a lipid membrane s…
Non-Hermitian second-order topological insulator with point gap
Xue-Min Yang, Hao Lin, Jian Li +3
The zero-mode corner states in the gap of two-dimensional non-Hermitian Su-Schrieffer-Heeger model are robust to infinitesimal perturbations that preserve chiral symmetry. However,…
Dirac-vortex topological cavity
Xiaomei Gao, Lechen Yang, Hao Lin +5
Cavity design is crucial for single-mode semiconductor lasers such as the distributed feedback (DFB) and vertical-cavity surface-emitting lasers (VCSEL). By recognizing that both o…
Temporal Analysis of Transaction Ego Networks with Different Labels on Ethereum
Baoying Huang, Jieli Liu, Jiajing Wu +2
Due to the widespread use of smart contracts, Ethereum has become the second-largest blockchain platform after Bitcoin. Many different types of Ethereum accounts (ICO, Mining, Gamb…
Dynamic Optimization and Safety Indicator Injection for Jailbreaking Text-to-Image Models with Multimodal Safety Filters
Zixuan Chen, Hao Lin, Ke Xu +2
Text-to-image (T2I) models can generate not-safe-for-work (NSFW) content, motivating multi-stage safety pipelines with both text and image filters. Newer LLM-based filters detect l…
DynaTrain: Fast Online Parallelism Switching for Elastic LLM Training
Yuanqing Wang, Yuchen Zhang, Hao Lin +9
Modern large language model (LLM) training is inherently dynamic: resource fluctuations, RLHF phase shifts, and cluster elasticity continually reshape the optimal parallelism layou…
Focused Forcing: Content-Aware Per-Frame KV Selection for Efficient Autoregressive Video Diffusion
Peiliang Cai, Evelyn Zhang, Jiacheng Liu +8
Recent advances in autoregressive video diffusion have enabled sequential and streaming video generation. However, long-horizon generation requires increasingly large KV caches, ma…
WoVR: World Models as Reliable Simulators for Post-Training VLA Policies with RL
Zhennan Jiang, Shangqing Zhou, Yutong Jiang +11
Reinforcement learning (RL) promises to unlock capabilities beyond imitation learning for Vision--Language--Action (VLA) models, but its requirement for massive real-world interact…
UniAP: Unifying Inter- and Intra-Layer Automatic Parallelism by Mixed Integer Quadratic Programming
Hao Lin, Ke Wu, Jie Li +2
Distributed learning is commonly used for training deep learning models, especially large models. In distributed learning, manual parallelism (MP) methods demand considerable human…
Uniform Turán densities of -uniform hypergraphs
Hao Lin, Guowei Sun, Guanghui Wang +1
For , the -uniform Turán density of a -graph is the supremum of for which there are arbitrarily large -free -graphs that are uniforml…
Domain-Aware RAG: MoL-Enhanced RL for Efficient Training and Scalable Retrieval
Hao Lin, Peitong Xie, Jingxue Chen +3
Retrieval-Augmented Generation (RAG) systems rely heavily on the retrieval stage, particularly the coarse-ranking process. Existing coarse-ranking optimization approaches often str…
Memory Layer: Train the In-Model Cache for Recommendation Models
Liangyuan Na, Gufan Yin, Yixin Bao +19
Early ranking stages in recommendation systems precompute item embeddings and cache them in-model for scoring within strict latency constraints. Because this cache exists only at s…
Stabilizing Reinforcement Learning with LLMs: Formulation and Practices
Chujie Zheng, Kai Dang, Bowen Yu +10
This paper proposes a novel formulation for reinforcement learning (RL) with large language models, explaining why and under what conditions the true sequence-level reward can be o…
Integer colorings with no rainbow -term arithmetic progression
Hao Lin, Guanghui Wang, Wenling Zhou
In this paper, we study the rainbow ErdÅs-Rothschild problem with respect to -term arithmetic progressions. For a set of positive integers , an -coloring of…
Investigating the completeness and omission roads of OpenStreetMap data in Hubei, China by comparing with Street Map and Street View
Qi Zhou, Hao Lin
OpenStreetMap (OSM) is a free map of the world which can be edited by global volunteers. Existing studies have showed that completeness of OSM road data in some developing countrie…
The minimum positive uniform Turán density in uniformly dense -uniform hypergraphs
Hao Lin, Guanghui Wang, Wenling Zhou
A -graph (or -uniform hypergraph) is uniformly dense if the edge distribution of is uniformly dense with respect to every large collection of -vertex cliques induc…