arXiv subjects

Le Zhang

Publications and source records attributed to Le Zhang.

At least 73 records · Page 4Linked to original sources

arXiv2026

Legal-DC: Benchmarking Retrieval-Augmented Generation for Legal Documents

Retrieval-Augmented Generation (RAG) has emerged as a promising technology for legal document consultation, yet its application in Chinese legal scenarios faces two key limitations: existing benchmarks lack specialized support for joint retriever-generator evaluation, and mainstream RAG systems often fail to accommodate the structured nature of legal provisions. To address these gaps, this study advances two core contributions: First, we constructed the Legal-DC benchmark dataset, comprising 480 legal documents (covering areas such as market regulation and contract management) and 2,475 refined question-answer pairs, each annotated with clause-level references, filling the gap for specialized evaluation resources in Chinese legal RAG. Second, we propose the LegRAG framework, which integrates legal adaptive indexing (clause-boundary segmentation) with a dual-path self-reflection mechanism to ensure clause integrity while enhancing answer accuracy. Third, we introduce automated evaluation methods for large language models to meet the high-reliability demands of legal retrieval scenarios. LegRAG outperforms existing state-of-the-art methods by 1.3% to 5.6% across key evaluation metrics. This research provides a specialized benchmark, practical framework, and empirical insights to advance the development of Chinese legal RAG systems. Our code and data are available at https://github.com/legal-dc/Legal-DC.

cs.CL↗

arXiv2026

UltraUPConvNet: A UPerNet- and ConvNeXt-Based Multi-Task Network for Ultrasound Tissue Segmentation and Disease Prediction

Ultrasound imaging is widely used in clinical practice due to its cost-effectiveness, mobility, and safety. However, current AI research often treats disease prediction and tissue segmentation as two separate tasks and their model requires substantial computational overhead. In such a situation, we introduce UltraUPConvNet, a computationally efficient universal framework designed for both ultrasound image classification and segmentation. Trained on a large-scale dataset containing more than 9,700 annotations across seven different anatomical regions, our model achieves state-of-the-art performance on certain datasets with lower computational overhead. Our model weights and codes are available at https://github.com/yyxl123/UltraUPConvNet

eess.IV↗

arXiv2026

Towards Real Zero-Shot Camouflaged Object Segmentation without Camouflaged Annotations

Camouflaged Object Segmentation (COS) faces significant challenges due to the scarcity of annotated data, where meticulous pixel-level annotation is both labor-intensive and costly, primarily due to the intricate object-background boundaries. Addressing the core question, "Can COS be effectively achieved in a zero-shot manner without manual annotations for any camouflaged object?" we affirmatively respond and introduce a robust zero-shot COS framework. This framework leverages the inherent local pattern bias of COS and employs a broad semantic feature space derived from salient object segmentation (SOS) for efficient zero-shot transfer. We incorporate an Masked Image Modeling (MIM) based image encoder optimized for Parameter-Efficient Fine-Tuning (PEFT), a Multimodal Large Language Model (M-LLM), and a Multi-scale Fine-grained Alignment (MFA) mechanism. The MIM pre-trained image encoder focuses on capturing essential low-level features, while the M-LLM generates caption embeddings processed alongside these visual cues. These embeddings are precisely aligned using MFA, enabling our framework to accurately interpret and navigate complex semantic contexts. To optimize operational efficiency, we introduce a learnable codebook that represents the M-LLM during inference, significantly reducing computational overhead. Our framework demonstrates its versatility and efficacy through rigorous experimentation, achieving state-of-the-art performance in zero-shot COS with $F_β^w$ scores of 72.9\% on CAMO and 71.7\% on COD10K. By removing the M-LLM during inference, we achieve an inference speed comparable to that of traditional end-to-end models, reaching 18.1 FPS. Code: https://github.com/AVC2-UESTC/ZSCOS-CaMF

cs.CV↗

arXiv2026

Seeing the Forest and the Trees: Query-Aware Tokenizer for Long-Video Multimodal Language Models

Despite the recent advances in the video understanding ability of multimodal large language models (MLLMs), long video understanding remains a challenge. One of the main issues is that the number of vision tokens grows linearly with video length, which causes an explosion in attention cost, memory, and latency. To solve this challenge, we present Query-aware Token Selector (\textbf{QTSplus}), a lightweight yet powerful visual token selection module that serves as an information gate between the vision encoder and LLMs. Given a text query and video tokens, QTSplus dynamically selects the most important visual evidence for the input text query by (i) scoring visual tokens via cross-attention, (ii) \emph{predicting} an instance-specific retention budget based on the complexity of the query, and (iii) \emph{selecting} Top-$n$ tokens with a differentiable straight-through estimator during training and a hard gate at inference. Furthermore, a small re-encoder preserves temporal order using absolute time information, enabling second-level localization while maintaining global coverage. Integrated into Qwen2.5-VL, QTSplus compresses the vision stream by up to \textbf{89\%} and reduces end-to-end latency by \textbf{28\%} on long videos. The evaluation on eight long video understanding benchmarks shows near-parity accuracy overall when compared with the original Qwen models and outperforms the original model by \textbf{+20.5} and \textbf{+5.6} points respectively on TempCompass direction and order accuracies. These results show that QTSplus is an effective, general mechanism for scaling MLLMs to real-world long-video scenarios while preserving task-relevant evidence.

cs.CV↗

arXiv2026

AnimeAgent: Is the Multi-Agent via Image-to-Video models a Good Disney Storytelling Artist?

Custom Storyboard Generation (CSG) aims to produce high-quality, multi-character consistent storytelling. Current approaches based on static diffusion models, whether used in a one-shot manner or within multi-agent frameworks, face three key limitations: (1) Static models lack dynamic expressiveness and often resort to "copy-paste" pattern. (2) One-shot inference cannot iteratively correct missing attributes or poor prompt adherence. (3) Multi-agents rely on non-robust evaluators, ill-suited for assessing stylized, non-realistic animation. To address these, we propose AnimeAgent, the first Image-to-Video (I2V)-based multi-agent framework for CSG. Inspired by Disney's "Combination of Straight Ahead and Pose to Pose" workflow, AnimeAgent leverages I2V's implicit motion prior to enhance consistency and expressiveness, while a mixed subjective-objective reviewer enables reliable iterative refinement. We also collect a human-annotated CSG benchmark with ground-truth. Experiments show AnimeAgent achieves SOTA performance in consistency, prompt fidelity, and stylization.

cs.CV↗

arXiv2026

YOLOv10-Based Multi-Task Framework for Hand Localization and Laterality Classification in Surgical Videos

Real-time hand tracking in trauma surgery is essential for supporting rapid and precise intraoperative decisions. We propose a YOLOv10-based framework that simultaneously localizes hands and classifies their laterality (left or right) in complex surgical scenes. The model is trained on the Trauma THOMPSON Challenge 2025 Task 2 dataset, consisting of first-person surgical videos with annotated hand bounding boxes. Extensive data augmentation and a multi-task detection design improve robustness against motion blur, lighting variations, and diverse hand appearances. Evaluation demonstrates accurate left-hand (67\%) and right-hand (71\%) classification, while distinguishing hands from the background remains challenging. The model achieves an $mAP_{[0.5:0.95]}$ of 0.33 and maintains real-time inference, highlighting its potential for intraoperative deployment. This work establishes a foundation for advanced hand-instrument interaction analysis in emergency surgical procedures.

cs.CV↗

arXiv2026

Unrewarded Exploration in Large Language Models Reveals Latent Learning from Psychology

Latent learning, classically theorized by Tolman, shows that biological agents (e.g., rats) can acquire internal representations of their environment without rewards, enabling rapid adaptation once rewards are introduced. In contrast, from a cognitive science perspective, reward learning remains overly dependent on external feedback, limiting flexibility and generalization. Although recent advances in the reasoning capabilities of large language models (LLMs), such as OpenAI-o1 and DeepSeek-R1, mark a significant breakthrough, these models still rely primarily on reward-centric reinforcement learning paradigms. Whether and how the well-established phenomenon of latent learning in psychology can inform or emerge within LLMs' training remains largely unexplored. In this work, we present novel findings from our experiments that LLMs also exhibit the latent learning dynamics. During an initial phase of unrewarded exploration, LLMs display modest performance improvements, as this phase allows LLMs to organize task-relevant knowledge without being constrained by reward-driven biases, and performance is further enhanced once rewards are introduced. LLMs post-trained under this two-stage exploration regime ultimately achieve higher competence than those post-trained with reward-based reinforcement learning throughout. Beyond these empirical observations, we also provide theoretical analyses for our experiments explaining why unrewarded exploration yields performance gains, offering a mechanistic account of these dynamics. Specifically, we conducted extensive experiments across multiple model families and diverse task domains to establish the existence of the latent learning dynamics in LLMs.

cs.LG↗

arXiv2026

KID: Knowledge-Injected Dual-Head Learning for Knowledge-Grounded Harmful Meme Detection

Internet memes have become pervasive carriers of digital culture on social platforms. However, their heavy reliance on metaphors and sociocultural context also makes them subtle vehicles for harmful content, posing significant challenges for automated content moderation. Existing approaches primarily focus on intra-modal and inter-modal signal analysis, while the understanding of implicit toxicity often depends on background knowledge that is not explicitly present in the meme itself. To address this challenge, we propose KID, a Knowledge-Injected Dual-Head Learning framework for knowledge-grounded harmful meme detection. KID adopts a label-constrained distillation paradigm to decompose complex meme understanding into structured reasoning chains that explicitly link visual evidence, background knowledge, and classification labels. These chains guide the learning process by grounding external knowledge in meme-specific contexts. In addition, KID employs a dual-head architecture that jointly optimizes semantic generation and classification objectives, enabling aligned linguistic reasoning while maintaining stable decision boundaries. Extensive experiments on five multilingual datasets spanning English, Chinese, and low-resource Bengali demonstrate that KID achieves SOTA performance on both binary and multi-label harmful meme detection tasks, improving over previous best methods by 2.1%--19.7% across primary evaluation metrics. Ablation studies further confirm the effectiveness of knowledge injection and dual-head joint learning, highlighting their complementary contributions to robust and generalizable meme understanding. The code and data are available at https://github.com/PotatoDog1669/KID.

cs.CL↗

arXiv2026

OmegaUse: Building a General-Purpose GUI Agent for Autonomous Task Execution

Graphical User Interface (GUI) agents show great potential for enabling foundation models to complete real-world tasks, revolutionizing human-computer interaction and improving human productivity. In this report, we present OmegaUse, a general-purpose GUI agent model for autonomous task execution on both mobile and desktop platforms, supporting computer-use and phone-use scenarios. Building an effective GUI agent model relies on two factors: (1) high-quality data and (2) effective training methods. To address these, we introduce a carefully engineered data-construction pipeline and a decoupled training paradigm. For data construction, we leverage rigorously curated open-source datasets and introduce a novel automated synthesis framework that integrates bottom-up autonomous exploration with top-down taxonomy-guided generation to create high-fidelity synthetic data. For training, to better leverage these data, we adopt a two-stage strategy: Supervised Fine-Tuning (SFT) to establish fundamental interaction syntax, followed by Group Relative Policy Optimization (GRPO) to improve spatial grounding and sequential planning. To balance computational efficiency with agentic reasoning capacity, OmegaUse is built on a Mixture-of-Experts (MoE) backbone. To evaluate cross-terminal capabilities in an offline setting, we introduce OS-Nav, a benchmark suite spanning multiple operating systems: ChiM-Nav, targeting Chinese Android mobile environments, and Ubu-Nav, focusing on routine desktop interactions on Ubuntu. Extensive experiments show that OmegaUse is highly competitive across established GUI benchmarks, achieving a state-of-the-art (SOTA) score of 96.3% on ScreenSpot-V2 and a leading 79.1% step success rate on AndroidControl. OmegaUse also performs strongly on OS-Nav, reaching 74.24% step success on ChiM-Nav and 55.9% average success on Ubu-Nav.

cs.AI↗

arXiv2026

Trustworthy Longitudinal Brain MRI Completion: A Deformation-Based Approach with KAN-Enhanced Diffusion Model

Longitudinal brain MRI is essential for lifespan study, yet high attrition rates often lead to missing data, complicating analysis. Deep generative models have been explored, but most rely solely on image intensity, leading to two key limitations: 1) the fidelity or trustworthiness of the generated brain images are limited, making downstream studies questionable; 2) the usage flexibility is restricted due to fixed guidance rooted in the model structure, restricting full ability to versatile application scenarios. To address these challenges, we introduce DF-DiffCom, a Kolmogorov-Arnold Networks (KAN)-enhanced diffusion model that smartly leverages deformation fields for trustworthy longitudinal brain image completion. Trained on OASIS-3, DF-DiffCom outperforms state-of-the-art methods, improving PSNR by 5.6% and SSIM by 0.12. More importantly, its modality-agnostic nature allows smooth extension to varied MRI modalities, even to attribute maps such as brain tissue segmentation results.

cs.CV↗

arXiv2026

Spin-orbit-driven quarter semimetals in rhombohedral graphene

Semimetals exhibit intriguing characteristics attributed to the coexistence of both electrons and holes. In rhombohedral multilayer graphene, a strong trigonal warping effect gives rise to a semi-metallic state near the Fermi surface, offering unique opportunities to explore the interplay of semi-metallic properties with strong correlations and topologies. Here, the observation of quarter semimetals in rhombohedral multilayer graphene by introducing spin-orbit coupling (SOC) is reported. The semi-metallic characteristics of rhombohedral graphene manifest as nearly vanished Hall resistance and parabolic longitudinal resistance. The strong correlations arising from the surface flat band lead to spontaneous symmetry breaking. SOC proximitized by WSe2 further lifts the valley degeneracy, resulting in the spontaneous time-reversal symmetry breaking, as evidenced by the hysteretic anomalous Hall effect. The coexistence of fully polarized electrons and holes allows for the observation of a non-monotonic temperature dependence of the anomalous Hall resistance. Furthermore, the application of moderate magnetic fields induces a phase transition from quarter semimetals to Chern insulators. These findings establish rhombohedral multilayer graphene as an ideal platform for studying strong correlations and topologies in semimetals.

cond-mat.mes-hall↗

arXiv2025

Holistic Evaluation of State-of-the-Art LLMs for Code Generation

This study presents a comprehensive empirical evaluation of six state-of-the-art large language models (LLMs) for code generation, including both general-purpose and code-specialized models. Using a dataset of 944 real-world LeetCode problems across five programming languages, we assess model performance using rigorous metrics: compile-time errors, runtime errors, functional failures, and algorithmic suboptimalities. The results reveal significant performance variations, with DeepSeek-R1 and GPT-4.1 consistently outperform others in terms of correctness, efficiency, and robustness. Through detailed case studies, we identify common failure scenarios such as syntax errors, logical flaws, and suboptimal algorithms, highlighting the critical role of prompt engineering and human oversight in improving results. Based on these findings, we provide actionable recommendations for developers and practitioners, emphasizing that successful LLM deployment depends on careful model selection, effective prompt design, and context-aware usage to ensure reliable code generation in real-world software development tasks.

cs.SE↗

arXiv2025

Photo-$z$ Estimation with Normalizing Flow

Accurate photometric redshift (photo-$z$) estimation is a key challenge in cosmology, as uncertainties in photo-$z$ directly limit the scientific return of large-scale structure and weak lensing studies, especially in upcoming Stage IV surveys. The problem is particularly severe for faint galaxies with sparse spectroscopic training data. In this work, we introduce nflow-$z$, a novel photo-$z$ estimation method using the powerful machine learning technique of normalizing flow. nflow-$z$ explicitly models the redshift probability distribution conditioned on the observables such as fluxes and colors. We build two nflow-$z$ implementations, dubbed cINN and cNSF, and compare their performance. We demonstrate the effectiveness of nflow-$z$ on several datasets, including a CSST mock, the COSMOS2020 catalog, and samples from DES Y1, SDSS, and DESCaLS. Our evaluation against state-of-the-art algorithms shows that nflow-$z$ performs favorably. For instance, cNSF surpasses Random Forest, Multi-Layer Perceptron, and Convolutional Neutral Network on the CSST mock test. We also achieve a ~30% improvement over official results for the faint DESCaLS sample and outperform conditional Generative Adversarial Network and Mixture Density Network methods on the DES Y1 dataset test. Furthermore, nflow-$z$ is computationally efficient, requiring only a fraction of the computing time of some of the competing algorithms. Our algorithm is particularly effective for the faint sample with sparse training data, making it highly suitable for upcoming Stage IV surveys.

astro-ph.IM↗

arXiv2025

A New Wavelet Scattering Transform-Based Statistic for Cosmological Analysis of Large-Scale Structure

Large-scale structure (LSS) analysis in galaxy surveys is a powerful cosmological probe but is limited by tracer bias, which can obscure underlying information and weaken parameter constraints. Existing methods either model bias or restrict analyses to low-density regions, yet their sensitivity to bias remains poorly understood. We propose a novel method based on the wavelet scattering transform (WST) to distinguish LSS across cosmological models while mitigating tracer bias. Central to our approach are the WST $m$-mode ratios, $R^{\rm wst}$, a new statistical measure, and a high-density apodization preprocessing that smoothly rescales extreme values. We use a reduced chi-square to assess the cosmological parameter constraints and find that $R^{\rm wst}$, in the scale range $j \in [3,7]$, achieves $χ^2_{ν, \rm cos} \approx 6$ for cosmology while maintaining $χ^2_{ν, \rm bias} \sim 1$--a regime unattained by other statistics. $R^{\rm wst}$ thus provides robust cosmological sensitivity with effective bias mitigation for future surveys.

astro-ph.CO↗

arXiv2025

Enhancing Generalization of Depth Estimation Foundation Model via Weakly-Supervised Adaptation with Regularization

The emergence of foundation models has substantially advanced zero-shot generalization in monocular depth estimation (MDE), as exemplified by the Depth Anything series. However, given access to some data from downstream tasks, a natural question arises: can the performance of these models be further improved? To this end, we propose WeSTAR, a parameter-efficient framework that performs Weakly supervised Self-Training Adaptation with Regularization, designed to enhance the robustness of MDE foundation models in unseen and diverse domains. We first adopt a dense self-training objective as the primary source of structural self-supervision. To further improve robustness, we introduce semantically-aware hierarchical normalization, which exploits instance-level segmentation maps to perform more stable and multi-scale structural normalization. Beyond dense supervision, we introduce a cost-efficient weak supervision in the form of pairwise ordinal depth annotations to further guide the adaptation process, which enforces informative ordinal constraints to mitigate local topological errors. Finally, a weight regularization loss is employed to anchor the LoRA updates, ensuring training stability and preserving the model's generalizable knowledge. Extensive experiments on both realistic and corrupted out-of-distribution datasets under diverse and challenging scenarios demonstrate that WeSTAR consistently improves generalization and achieves state-of-the-art performance across a wide range of benchmarks.

cs.CV↗

arXiv2025

FunOTTA: On-the-Fly Adaptation on Cross-Domain Fundus Image via Stable Test-time Training

Fundus images are essential for the early screening and detection of eye diseases. While deep learning models using fundus images have significantly advanced the diagnosis of multiple eye diseases, variations in images from different imaging devices and locations (known as domain shifts) pose challenges for deploying pre-trained models in real-world applications. To address this, we propose a novel Fundus On-the-fly Test-Time Adaptation (FunOTTA) framework that effectively generalizes a fundus image diagnosis model to unseen environments, even under strong domain shifts. FunOTTA stands out for its stable adaptation process by performing dynamic disambiguation in the memory bank while minimizing harmful prior knowledge bias. We also introduce a new training objective during adaptation that enables the classifier to incrementally adapt to target patterns with reliable class conditional estimation and consistency regularization. We compare our method with several state-of-the-art test-time adaptation (TTA) pipelines. Experiments on cross-domain fundus image benchmarks across two diseases demonstrate the superiority of the overall framework and individual components under different backbone networks. Code is available at https://github.com/Casperqian/FunOTTA.

cs.CV↗

arXiv2025

Testing the Equivalence Principle on Cosmological Scales Using Peculiar Acceleration Power Spectrum

While the (weak) Equivalence Principle (EP) has been rigorously tested within the solar system, its validity on cosmological scales, particularly in the context of dark matter and dark energy, remains uncertain. In this study, we propose a novel method to test EP on cosmological scales by measuring the peculiar acceleration power spectrum of galaxies using the redshift drift technique. We develop an EP estimator, $E_{\rm ep}$, to evaluate the consistency of the peculiar acceleration power spectrum across different tracers. By calculating the ratio of the peculiar acceleration power spectra of tracers, the ensemble average of $E_{\rm ep}$ is expected to be unity if EP holds on cosmological scales for these tracers. We validate this estimator using N-body simulations, focusing on four redshift bins with $z\leq 1.5$ and scales of $k$ in the range of $0.007$ and $0.2$ $h/\rm Mpc$. By fitting a single parameter $δ_{\rm ep}$ across redshifts, we find that DM particle mocks without EP violation yield $δ_{\rm ep}$ consistent with zero under the small redshift measurement uncertainty case, while the large redshift uncertainty case slightly induces biases at low redshifts. In addition, when using DM halo mocks with controlled EP violations, no-violation and mild-violation cases show no significant detection, while moderate and strong violations produce statistically significant $δ_{\rm ep}$ values and high $χ^2$, especially at low redshifts, confirming the estimator's sensitivity. Taking advantage of advanced observing capabilities, such as next-generation facilities that extend beyond the Square Kilometer Array, the proposed method offers a promising approach for future cosmological tests of EP.

astro-ph.CO↗

arXiv2025

Prompt to Restore, Restore to Prompt: Cyclic Prompting for Universal Adverse Weather Removal

Universal adverse weather removal (UAWR) seeks to address various weather degradations within a unified framework. Recent methods are inspired by prompt learning using pre-trained vision-language models (e.g., CLIP), leveraging degradation-aware prompts to facilitate weather-free image restoration, yielding significant improvements. In this work, we propose CyclicPrompt, an innovative cyclic prompt approach designed to enhance the effectiveness, adaptability, and generalizability of UAWR. CyclicPrompt Comprises two key components: 1) a composite context prompt that integrates weather-related information and context-aware representations into the network to guide restoration. This prompt differs from previous methods by marrying learnable input-conditional vectors with weather-specific knowledge, thereby improving adaptability across various degradations. 2) The erase-and-paste mechanism, after the initial guided restoration, substitutes weather-specific knowledge with constrained restoration priors, inducing high-quality weather-free concepts into the composite prompt to further fine-tune the restoration process. Therefore, we can form a cyclic "Prompt-Restore-Prompt" pipeline that adeptly harnesses weather-specific knowledge, textual contexts, and reliable textures. Extensive experiments on synthetic and real-world datasets validate the superior performance of CyclicPrompt. The code is available at: https://github.com/RongxinL/CyclicPrompt.

cs.CV↗