arXiv subjects

Zhe Li

Publications and source records attributed to Zhe Li.

At least 109 records · Page 6Linked to original sources

arXiv2025

Commercial Vehicle Braking Optimization: A Robust SIFT-Trajectory Approach

A vision-based trajectory analysis solution is proposed to address the "zero-speed braking" issue caused by inaccurate Controller Area Network (CAN) signals in commercial vehicle Automatic Emergency Braking (AEB) systems during low-speed operation. The algorithm utilizes the NVIDIA Jetson AGX Xavier platform to process sequential video frames from a blind spot camera, employing self-adaptive Contrast Limited Adaptive Histogram Equalization (CLAHE)-enhanced Scale-Invariant Feature Transform (SIFT) feature extraction and K-Nearest Neighbors (KNN)-Random Sample Consensus (RANSAC) matching. This allows for precise classification of the vehicle's motion state (static, vibration, moving). Key innovations include 1) multiframe trajectory displacement statistics (5-frame sliding window), 2) a dual-threshold state decision matrix, and 3) OBD-II driven dynamic Region of Interest (ROI) configuration. The system effectively suppresses environmental interference and false detection of dynamic objects, directly addressing the challenge of low-speed false activation in commercial vehicle safety systems. Evaluation in a real-world dataset (32,454 video segments from 1,852 vehicles) demonstrates an F1-score of 99.96% for static detection, 97.78% for moving state recognition, and a processing delay of 14.2 milliseconds (resolution 704x576). The deployment on-site shows an 89% reduction in false braking events, a 100% success rate in emergency braking, and a fault rate below 5%.

cs.CV↗

arXiv2025

DeX-Portrait: Disentangled and Expressive Portrait Animation via Explicit and Latent Motion Representations

Portrait animation from a single source image and a driving video is a long-standing problem. Recent approaches tend to adopt diffusion-based image/video generation models for realistic and expressive animation. However, none of these diffusion models realizes high-fidelity disentangled control between the head pose and facial expression, hindering applications like expression-only or pose-only editing and animation. To address this, we propose DeX-Portrait, a novel approach capable of generating expressive portrait animation driven by disentangled pose and expression signals. Specifically, we represent the pose as an explicit global transformation and the expression as an implicit latent code. First, we design a powerful motion trainer to learn both pose and expression encoders for extracting precise and decomposed driving signals. Then we propose to inject the pose transformation into the diffusion model through a dual-branch conditioning mechanism, and the expression latent through cross attention. Finally, we design a progressive hybrid classifier-free guidance for more faithful identity consistency. Experiments show that our method outperforms state-of-the-art baselines on both animation quality and disentangled controllability.

cs.CV↗

arXiv2025

InfoMotion: A Graph-Based Approach to Video Dataset Distillation for Echocardiography

Echocardiography plays a critical role in the diagnosis and monitoring of cardiovascular diseases as a non-invasive real-time assessment of cardiac structure and function. However, the growing scale of echocardiographic video data presents significant challenges in terms of storage, computation, and model training efficiency. Dataset distillation offers a promising solution by synthesizing a compact, informative subset of data that retains the key clinical features of the original dataset. In this work, we propose a novel approach for distilling a compact synthetic echocardiographic video dataset. Our method leverages motion feature extraction to capture temporal dynamics, followed by class-wise graph construction and representative sample selection using the Infomap algorithm. This enables us to select a diverse and informative subset of synthetic videos that preserves the essential characteristics of the original dataset. We evaluate our approach on the EchoNet-Dynamic datasets and achieve a test accuracy of $69.38\%$ using only $25$ synthetic videos. These results demonstrate the effectiveness and scalability of our method for medical video dataset distillation.

cs.CV↗

arXiv2025

Label-free Motion-Conditioned Diffusion Model for Cardiac Ultrasound Synthesis

Ultrasound echocardiography is essential for the non-invasive, real-time assessment of cardiac function, but the scarcity of labelled data, driven by privacy restrictions and the complexity of expert annotation, remains a major obstacle for deep learning methods. We propose the Motion Conditioned Diffusion Model (MCDM), a label-free latent diffusion framework that synthesises realistic echocardiography videos conditioned on self-supervised motion features. To extract these features, we design the Motion and Appearance Feature Extractor (MAFE), which disentangles motion and appearance representations from videos. Feature learning is further enhanced by two auxiliary objectives: a re-identification loss guided by pseudo appearance features and an optical flow loss guided by pseudo flow fields. Evaluated on the EchoNet-Dynamic dataset, MCDM achieves competitive video generation performance, producing temporally coherent and clinically realistic sequences without reliance on manual labels. These results demonstrate the potential of self-supervised conditioning for scalable echocardiography synthesis. Our code is available at https://github.com/ZheLi2020/LabelfreeMCDM.

cs.CV↗

arXiv2025

Video Dataset Condensation with Diffusion Models

In recent years, the rapid expansion of dataset sizes and the increasing complexity of deep learning models have significantly escalated the demand for computational resources, both for data storage and model training. Dataset distillation has emerged as a promising solution to address this challenge by generating a compact synthetic dataset that retains the essential information from a large real dataset. However, existing methods often suffer from limited performance, particularly in the video domain. In this paper, we focus on video dataset distillation. We begin by employing a video diffusion model to generate synthetic videos. Since the videos are generated only once, this significantly reduces computational costs. Next, we introduce the Video Spatio-Temporal U-Net (VST-UNet), a model designed to select a diverse and informative subset of videos that effectively captures the characteristics of the original dataset. To further optimize computational efficiency, we explore a training-free clustering algorithm, Temporal-Aware Cluster-based Distillation (TAC-DT), to select representative videos without requiring additional training overhead. We validate the effectiveness of our approach through extensive experiments on four benchmark datasets, demonstrating performance improvements of up to $10.61\%$ over the state-of-the-art. Our method consistently outperforms existing approaches across all datasets, establishing a new benchmark for video dataset distillation.

cs.CV↗

arXiv2025

Leveraging Multi-Modal Information to Enhance Dataset Distillation

Dataset distillation aims to create a small and highly representative synthetic dataset that preserves the essential information of a larger real dataset. Beyond reducing storage and computational costs, related approaches offer a promising avenue for privacy preservation in computer vision by eliminating the need to store or share sensitive real-world images. Existing methods focus solely on optimizing visual representations, overlooking the potential of multi-modal information. In this work, we propose a multi-modal dataset distillation framework that incorporates two key enhancements: caption-guided supervision and object-centric masking. To leverage textual information, we introduce two strategies: caption concatenation, which fuses caption embeddings with visual features during classification, and caption matching, which enforces semantic alignment between real and synthetic data through a caption-based loss. To improve data utility and reduce unnecessary background noise, we employ segmentation masks to isolate target objects and introduce two novel losses: masked feature alignment and masked gradient matching, both aimed at promoting object-centric learning. Extensive evaluations demonstrate that our approach improves downstream performance while promoting privacy protection by minimizing exposure to real data.

cs.CV↗

arXiv2025

SoK: a Comprehensive Causality Analysis Framework for Large Language Model Security

Large Language Models (LLMs) exhibit remarkable capabilities but remain vulnerable to adversarial manipulations such as jailbreaking, where crafted prompts bypass safety mechanisms. Understanding the causal factors behind such vulnerabilities is essential for building reliable defenses. In this work, we introduce a unified causality analysis framework that systematically supports all levels of causal investigation in LLMs, ranging from token-level, neuron-level, and layer-level interventions to representation-level analysis. The framework enables consistent experimentation and comparison across diverse causality-based attack and defense methods. Accompanying this implementation, we provide the first comprehensive survey of causality-driven jailbreak studies and empirically evaluate the framework on multiple open-weight models and safety-critical benchmarks including jailbreaks, hallucination detection, backdoor identification, and fairness evaluation. Our results reveal that: (1) targeted interventions on causally critical components can reliably modify safety behavior; (2) safety-related mechanisms are highly localized (i.e., concentrated in early-to-middle layers with only 1--2\% of neurons exhibiting causal influence); and (3) causal features extracted from our framework achieve over 95\% detection accuracy across multiple threat types. By bridging theoretical causality analysis and practical model safety, our framework establishes a reproducible foundation for research on causality-based attacks, interpretability, and robust attack detection and mitigation in LLMs. Code is available at https://github.com/Amadeuszhao/SOK_Casuality.

cs.CR↗

arXiv2025

Singular jets during droplet impact on superhydrophobic surfaces

Hypothesis: The impact of droplets is prevalent in numerous applications, and jetting during droplet impact is a critical process controlling the dispersal and transport of liquid. New jetting dynamics are expected in different conditions of droplet impact on super-hydrophobic surfaces, such as new jetting phenomena, mechanisms, and regimes. Experiments: In this experimental study of droplet impact on super-hydrophobic surfaces, the Weber number and the Ohnesorge number are varied in a wide range, and the impact process is analyzed theoretically. Findings: We identify a new type of singular jets, i.e., singular jets induced by horizontal inertia (HI singular jets), besides the previously studied singular jets induced by capillary deformation (CD singular jets). For CD singular jets, the formation of the cavity is due to the propagation of capillary waves on the droplet surface; while for HI singular jets, the cavity formation is due to the large horizontal inertia of the toroidal edge during the retraction of the droplet after the maximum spreading. Key steps of the impact process are analyzed quantitatively, including the spreading of the droplet, the formation and the collapse of the spire, the formation and retraction of the cavity, and finally the formation of singular jets. A regime map for the formation of singular jets is obtained, and scaling relationships for the transition conditions between different regimes are analyzed.

physics.flu-dyn↗

arXiv2025

KRAL: Knowledge and Reasoning Augmented Learning for LLM-assisted Clinical Antimicrobial Therapy

Clinical antimicrobial therapy requires the dynamic integration of pathogen profiles,host factors, pharmacological properties of antimicrobials,and the severity of infection. This complexity imposes fundamental limitations on the applicability of Large Language Models (LLMs) in high-stakes clinical decision-making including knowledge gaps, data privacy concerns, high deployment costs, and limited reasoning capabilities. To address these challenges, we propose KRAL (Knowledge and Reasoning Augmented Learning), a low-cost, scalable, privacy-preserving paradigm that leverages teacher-model reasoning to automatically distill knowledge and reasoning trajectories via answer-to-question reverse generation, employs heuristic learning for semi-supervised data augmentation (reducing manual annotation requirements by approximately 80%), and utilizes agentic reinforcement learning to jointly enhance medical knowledge and reasoning while optimizing computational and memory efficiency. A hierarchical evaluation employing diverse teacher-model proxies reduces assessment costs, while modular interface design facilitates seamless system updates. Experimental results demonstrate that KRAL significantly outperforms traditional Retrieval-Augmented Generation (RAG) and Supervised Fine-Tuning (SFT) methods. It improves knowledge question-answering capability (Accuracy@1 on the external open-source benchmark MEDQA increased by 1.8% vs. SFT and 3.6% vs. RAG) and reasoning capability (Pass@1 on the external benchmark PUMCH Antimicrobial increased by 27% vs. SFT and 27.2% vs. RAG), achieved at about 20% of SFT's long-term training costs. This establishes KRAL as an effective solution for enhancing local LLMs' clinical diagnostic capabilities, enabling low-cost, high-safety deployment in complex medical decision support.

cs.AI↗

arXiv2025

Floquet-engineered Valley Topology with Anisotropic Response in 1T'-WSe$_2$ and Janus WSeTe monolayers

Valley topology has emerged as a key concept for realizing new classes of quantum states. Here, we investigate Floquet-engineered topological phase transitions in anisotropic 1T'-WSe$_2$ and its Janus derivative WSeTe monolayers, which exhibit valley-degenerate and valley-polarized characteristics, respectively. In 1T'-WSe$_2$, a single topological-phase-transition (TPT) occurs from the quantum-spin-Hall state (QSH) to the quantum anomalous Hall (QAH) state, involving one spin channel at both valleys simultaneously. In contrast, Janus WSeTe undergoes a two-stage Floquet-driven TPT that occurs within a single valley and sequentially involves two spin components. The intermediate phase manifests as a valley-polarized QAH (vp-QAH) state with a finite valley Chern number, while the final phase evolves into a high-Chern-number QAH state with distinct valley gaps. Furthermore, an in-plane anisotropic response of the TPTs is predicted under oblique light incidence, reflecting the intrinsic low-symmetry nature of the lattice. These findings provide a comprehensive understanding of Floquet-engineered valley-based topological properties and offer guidance for designing light-controllable valleytronic and topological devices.

cond-mat.mes-hall↗

arXiv2025

Light-engineered Multichannel Quantum Anomalous Hall Effect in High-order Topological Plumbene

Floquet engineering severs as a forceful technique for uncovering high Chern numbers of quantum anomalous Hall (QAH) states with feasible tunability in high-order topologically insulating plumbene, which is readily accessible for experimental investigations. Under the irradiation of righthanded circularly polarized light, we predict a three-stage topological phase transition in plumbene, whether it is in a free-standing form or grown on h-BN. Initially, a metallic state evolves into a K(K')-valley-based QAH state with a Chern number of -8, which then decreases to -6 after the valley gap closes. Finally, a band inversion occurs at the $Γ$ point, resulting in a multichannel QAH state with C = -3. The trigonal warping model accounts for both K(K')-valley-based and $Γ$-pointbased QAH states. Additionally, growing plumbene on a non-van-der-Waals substrate eliminates the K(K')-valley-based topology, leaving only the $Γ$-point-based QAH state with C = +3. Our findings propose the tunability of various high Chern numbers derived from high-order topological insulators, aiming to advance the next-generation dissipationless electronic devices.

cond-mat.mes-hall↗

arXiv2025

Q-MLLM: Vector Quantization for Robust Multimodal Large Language Model Security

Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities in cross-modal understanding, but remain vulnerable to adversarial attacks through visual inputs despite robust textual safety mechanisms. These vulnerabilities arise from two core weaknesses: the continuous nature of visual representations, which allows for gradient-based attacks, and the inadequate transfer of text-based safety mechanisms to visual content. We introduce Q-MLLM, a novel architecture that integrates two-level vector quantization to create a discrete bottleneck against adversarial attacks while preserving multimodal reasoning capabilities. By discretizing visual representations at both pixel-patch and semantic levels, Q-MLLM blocks attack pathways and bridges the cross-modal safety alignment gap. Our two-stage training methodology ensures robust learning while maintaining model utility. Experiments demonstrate that Q-MLLM achieves significantly better defense success rate against both jailbreak attacks and toxic image attacks than existing approaches. Notably, Q-MLLM achieves perfect defense success rate (100\%) against jailbreak attacks except in one arguable case, while maintaining competitive performance on multiple utility benchmarks with minimal inference overhead. This work establishes vector quantization as an effective defense mechanism for secure multimodal AI systems without requiring expensive safety-specific fine-tuning or detection overhead. Code is available at https://github.com/Amadeuszhao/QMLLM.

cs.CR↗

arXiv2025

AutoBackdoor: Automating Backdoor Attacks via LLM Agents

Backdoor attacks pose a serious threat to the secure deployment of large language models (LLMs), enabling adversaries to implant hidden behaviors triggered by specific inputs. However, existing methods often rely on manually crafted triggers and static data pipelines, which are rigid, labor-intensive, and inadequate for systematically evaluating modern defense robustness. As AI agents become increasingly capable, there is a growing need for more rigorous, diverse, and scalable \textit{red-teaming frameworks} that can realistically simulate backdoor threats and assess model resilience under adversarial conditions. In this work, we introduce \textsc{AutoBackdoor}, a general framework for automating backdoor injection, encompassing trigger generation, poisoned data construction, and model fine-tuning via an autonomous agent-driven pipeline. Unlike prior approaches, AutoBackdoor uses a powerful language model agent to generate semantically coherent, context-aware trigger phrases, enabling scalable poisoning across arbitrary topics with minimal human effort. We evaluate AutoBackdoor under three realistic threat scenarios, including \textit{Bias Recommendation}, \textit{Hallucination Injection}, and \textit{Peer Review Manipulation}, to simulate a broad range of attacks. Experiments on both open-source and commercial models, including LLaMA-3, Mistral, Qwen, and GPT-4o, demonstrate that our method achieves over 90\% attack success with only a small number of poisoned samples. More importantly, we find that existing defenses often fail to mitigate these attacks, underscoring the need for more rigorous and adaptive evaluation techniques against agent-driven threats as explored in this work. All code, datasets, and experimental configurations will be merged into our primary repository at https://github.com/bboylyg/BackdoorLLM.

cs.CR↗

arXiv2025

Graph-Based Imitation and Reinforcement Learning for Efficient Benders Decomposition

This work introduces an end-to-end graph-based agent for accelerating the computational efficiency of Benders Decomposition. The agent's policy is parameterized by a graph neural network which takes as input a bipartite graph representation of the master problem and proposes a candidate solution. The agent is trained using a two-stage approach that combines imitation (IL) and reinforcement learning (RL). IL is used to mimic a solver and obtain a warm-start policy which is then finetuned using RL with a reward signal that balances feasibility and computational efficiency. We augment the agent with a verification mechanism that checks the agent's prediction for feasibility and solution quality. The framework is evaluated in two case studies: (i) an illustrative mixed-integer nonlinear program, where it reduces the solution time by 42% without loss of solution quality, and (ii) a closed-loop irrigation scheduling problem, where it achieves a 23% time reduction without compromising water use efficiency.

math.OC↗

arXiv2025

Agentar-DeepFinance-100K: A Large-Scale Financial Dataset via Systematic Chain-of-Thought Synthesis Optimization

Recent advancements in large language models (LLMs) have demonstrated remarkable general reasoning capabilities, holding significant potential for applications in the financial domain, a field that requires robust and reliable reasoning. It has been demonstrated that distilling high-quality chain-of-thought (CoT) rationales from advanced general reasoning models offers a promising and efficient path to the financial reasoning model. However, existing CoT synthesis methods suffer from shallow CoT sampling, leaving the question of how to construct a well-designed knowledge space for finance reasoning unexplored. In this paper, we present Agentar-DeepFinance-100K, a large-scale financial reasoning dataset characterized by its systematic CoT synthesis optimization. We first introduce a comprehensive CoT synthesis pipeline featuring Multi-perspective Knowledge Extraction (MKE) and Self-Corrective Rewriting (SCR) to generate exhaustive and deep financial reasoning trajectories. Furthermore, a systematic investigation, termed CoT Cube, is conducted to analyze critical factors that influence CoT effectiveness, such as necessity, length and synthesizer, yielding valuable insights for high-quality financial CoT construction. Experiments demonstrate that models trained on our Agentar-DeepFinance-100K achieve significant improvements on financial benchmarks. We publicly release Agentar-DeepFinance-100K , hoping to advance the research in financial reasoning models.

cs.CE↗

arXiv2025

URDF-Anything: Constructing Articulated Objects with 3D Multimodal Language Model

Constructing accurate digital twins of articulated objects is essential for robotic simulation training and embodied AI world model building, yet historically requires painstaking manual modeling or multi-stage pipelines. In this work, we propose \textbf{URDF-Anything}, an end-to-end automatic reconstruction framework based on a 3D multimodal large language model (MLLM). URDF-Anything utilizes an autoregressive prediction framework based on point-cloud and text multimodal input to jointly optimize geometric segmentation and kinematic parameter prediction. It implements a specialized $[SEG]$ token mechanism that interacts directly with point cloud features, enabling fine-grained part-level segmentation while maintaining consistency with the kinematic parameter predictions. Experiments on both simulated and real-world datasets demonstrate that our method significantly outperforms existing approaches regarding geometric segmentation (mIoU 17\% improvement), kinematic parameter prediction (average error reduction of 29\%), and physical executability (surpassing baselines by 50\%). Notably, our method exhibits excellent generalization ability, performing well even on objects outside the training set. This work provides an efficient solution for constructing digital twins for robotic simulation, significantly enhancing the sim-to-real transfer capability.

cs.RO↗

arXiv2025

EmbeddingGemma: Powerful and Lightweight Text Representations

We introduce EmbeddingGemma, a new lightweight, open text embedding model based on the Gemma 3 language model family. Our innovative training recipe strategically captures knowledge from larger models via encoder-decoder initialization and geometric embedding distillation. We improve model robustness and expressiveness with a spread-out regularizer, and ensure generalizability by merging checkpoints from varied, optimized mixtures. Evaluated on the Massive Text Embedding Benchmark (MTEB) across multilingual, English, and code domains, EmbeddingGemma (300M) achieves state-of-the-art results. Notably, it outperforms prior top models, both proprietary and open, with fewer than 500M parameters, and provides performance comparable to models double its size, offering an exceptional performance-to-cost ratio. Remarkably, this lead persists when quantizing model weights or truncating embedding outputs. This makes EmbeddingGemma particularly well-suited for low-latency and high-throughput use cases such as on-device applications. We provide ablation studies exploring our key design choices. We release EmbeddingGemma to the community to promote further research.

cs.CL↗

arXiv2025

Lightweight Ac Arc Fault Diagnosis via Fourier Transform Inspired Multi-frequency Neural Network

Lightweight online detection of series arc faults is critically needed in residential and industrial power systems to prevent electrical fires. Existing diagnostic methods struggle to achieve both rapid response and robust accuracy under resource-constrained conditions. To overcome the challenge, this work suggests leveraging a multi-frequency neural network named MFNN, embedding prior physical knowledge into the network. Inspired by arcing current curve and the Fourier decomposition analysis, we create an adaptive activation function with super-expressiveness, termed EAS, and a novel network architecture with branch networks to help MFNN extract features with multiple frequencies. In our experiments, eight advanced arc fault diagnosis models across an experimental dataset with multiple sampling times and multi-level noise are used to demonstrate the superiority of MFNN. The corresponding experiments show: 1) The MFNN outperforms other models in arc fault location, befitting from signal decomposition of branch networks. 2) The noise immunity of MFNN is much better than that of other models, achieving 14.51% over LCNN and 16.3% over BLS in test accuracy when SNR=-9. 3) EAS and the network architecture contribute to the excellent performance of MFNN.

eess.SP↗