arXiv subjects

Xu Wang

Publications and source records attributed to Xu Wang.

At least 163 records · Page 9Linked to original sources

arXiv2025

Q-Doc: Benchmarking Document Image Quality Assessment Capabilities in Multi-modal Large Language Models

The rapid advancement of Multi-modal Large Language Models (MLLMs) has expanded their capabilities beyond high-level vision tasks. Nevertheless, their potential for Document Image Quality Assessment (DIQA) remains underexplored. To bridge this gap, we propose Q-Doc, a three-tiered evaluation framework for systematically probing DIQA capabilities of MLLMs at coarse, middle, and fine granularity levels. a) At the coarse level, we instruct MLLMs to assign quality scores to document images and analyze their correlation with Quality Annotations. b) At the middle level, we design distortion-type identification tasks, including single-choice and multi-choice tests for multi-distortion scenarios. c) At the fine level, we introduce distortion-severity assessment where MLLMs classify distortion intensity against human-annotated references. Our evaluation demonstrates that while MLLMs possess nascent DIQA abilities, they exhibit critical limitations: inconsistent scoring, distortion misidentification, and severity misjudgment. Significantly, we show that Chain-of-Thought (CoT) prompting substantially enhances performance across all levels. Our work provides a benchmark for DIQA capabilities in MLLMs, revealing pronounced deficiencies in their quality perception and promising pathways for enhancement. The benchmark and code are publicly available at: https://github.com/cydxf/Q-Doc.

cs.CV↗

arXiv2025

DetailFlow: 1D Coarse-to-Fine Autoregressive Image Generation via Next-Detail Prediction

This paper presents DetailFlow, a coarse-to-fine 1D autoregressive (AR) image generation method that models images through a novel next-detail prediction strategy. By learning a resolution-aware token sequence supervised with progressively degraded images, DetailFlow enables the generation process to start from the global structure and incrementally refine details. This coarse-to-fine 1D token sequence aligns well with the autoregressive inference mechanism, providing a more natural and efficient way for the AR model to generate complex visual content. Our compact 1D AR model achieves high-quality image synthesis with significantly fewer tokens than previous approaches, i.e. VAR/VQGAN. We further propose a parallel inference mechanism with self-correction that accelerates generation speed by approximately 8x while reducing accumulation sampling error inherent in teacher-forcing supervision. On the ImageNet 256x256 benchmark, our method achieves 2.96 gFID with 128 tokens, outperforming VAR (3.3 FID) and FlexVAR (3.05 FID), which both require 680 tokens in their AR models. Moreover, due to the significantly reduced token count and parallel inference mechanism, our method runs nearly 2x faster inference speed compared to VAR and FlexVAR. Extensive experimental results demonstrate DetailFlow's superior generation quality and efficiency compared to existing state-of-the-art methods.

cs.CV↗

arXiv2025

Asymptotic error distribution of numerical methods for parabolic SPDEs with multiplicative noise

This paper aims to investigate the asymptotic error distribution of several numerical methods for stochastic partial differential equations (SPDEs) with multiplicative noise. Firstly, we give the limit distribution of the normalized error process of the exponential Euler method in $\dot{H}^η$ for some $η>0$. A key finding is that the asymptotic error in distribution of the exponential Euler method is governed by a linear SPDE driven by infinitely many independent $Q$-Wiener processes. This characteristic represents a significant difference from numerical methods for both stochastic ordinary differential equations and SPDEs with additive noise. Secondly, as applications of the above result, we derive the asymptotic error distribution of a full discretization based on the temporal exponential Euler method and the spatial finite element method. As a concrete illustration, we provide the pointwise limit distribution of the normalized error process when the exponential Euler method is applied to a specific class of stochastic heat equations. Finally, by studying the asymptotic error of the spatial semi-discrete spectral Galerkin method, we demonstrate that the actual strong convergence speed of spatial semi-discrete numerical methods may be highly problem-dependent, rather than universally predictable.

math.NA↗

arXiv2025

3D Weakly Supervised Semantic Segmentation via Class-Aware and Geometry-Guided Pseudo-Label Refinement

3D weakly supervised semantic segmentation (3D WSSS) aims to achieve semantic segmentation by leveraging sparse or low-cost annotated data, significantly reducing reliance on dense point-wise annotations. Previous works mainly employ class activation maps or pre-trained vision-language models to address this challenge. However, the low quality of pseudo-labels and the insufficient exploitation of 3D geometric priors jointly create significant technical bottlenecks in developing high-performance 3D WSSS models. In this paper, we propose a simple yet effective 3D weakly supervised semantic segmentation method that integrates 3D geometric priors into a class-aware guidance mechanism to generate high-fidelity pseudo labels. Concretely, our designed methodology first employs Class-Aware Label Refinement module to generate more balanced and accurate pseudo labels for semantic categrories. This initial refinement stage focuses on enhancing label quality through category-specific optimization. Subsequently, the Geometry-Aware Label Refinement component is developed, which strategically integrates implicit 3D geometric constraints to effectively filter out low-confidence pseudo labels that fail to comply with geometric plausibility. Moreover, to address the challenge of extensive unlabeled regions, we propose a Label Update strategy that integrates Self-Training to propagate labels into these areas. This iterative process continuously enhances pseudo-label quality while expanding label coverage, ultimately fostering the development of high-performance 3D WSSS models. Comprehensive experimental validation reveals that our proposed methodology achieves state-of-the-art performance on both ScanNet and S3DIS benchmarks while demonstrating remarkable generalization capability in unsupervised settings, maintaining competitive accuracy through its robust design.

cs.CV↗

arXiv2025

MeTA-LoRA: Data-Efficient Multi-Task Fine-Tuning for Large Language Models

Low-Rank Adaptation (LoRA) has emerged as one of the most widely used parameter-efficient fine-tuning (PEFT) methods for adapting large language models (LLMs) to downstream tasks. While highly effective in single-task settings, it struggles to efficiently leverage inter-task knowledge in complex multi-task learning scenarios, often requiring substantial task-specific data to achieve optimal performance. To address this limitation, we introduce MeTA-LoRA, a two-stage optimization framework that significantly improves data efficiency in multi-task adaptation. In the first stage, task-specific LoRA adapters are learned using only a few samples from each involved dataset, enabling rapid adaptation without large-scale supervision. In the second stage, the shared LoRA adapter is updated by aggregating gradients from multiple tasks to promote knowledge transfer across tasks, further reducing data usage by leveraging common patterns. In both multi-task learning and multilingual learning scenarios, our method matches or surpasses the performance of traditional full-data LoRA fine-tuning approaches, while using significantly less task-specific data.

cs.CL↗

arXiv2025

Does higher interpretability imply better utility? A Pairwise Analysis on Sparse Autoencoders

Sparse Autoencoders (SAEs) are widely used to steer large language models (LLMs), based on the assumption that their interpretable features naturally enable effective model behavior steering. Yet, a fundamental question remains unanswered: does higher interpretability indeed imply better steering utility? To answer this question, we train 90 SAEs across three LLMs (Gemma-2-2B, Qwen-2.5-3B, Gemma-2-9B), spanning five architectures and six sparsity levels, and evaluate their interpretability and steering utility based on SAEBench (arXiv:2501.12345) and AxBench (arXiv:2502.23456) respectively, and perform a rank-agreement analysis via Kendall's rank coefficients (tau b). Our analysis reveals only a relatively weak positive association (tau b approx 0.298), indicating that interpretability is an insufficient proxy for steering performance. We conjecture the interpretability utility gap may stem from the selection of SAE features, as not all of them are equally effective for steering. To further find features that truly steer the behavior of LLMs, we propose a novel selection criterion called Delta Token Confidence, which measures how much amplifying a feature changes the next token distribution. We show that our method improves the steering performance of three LLMs by 52.52 percent compared to the current best output score based criterion (arXiv:2503.34567). Strikingly, after selecting features with high Delta Token Confidence, the correlation between interpretability and utility vanishes (tau b approx 0), and can even become negative. This further highlights the divergence between interpretability and utility for the most effective steering features.

cs.LG↗

arXiv2025

BlockFUL: Enabling Unlearning in Blockchained Federated Learning

Unlearning in Federated Learning (FL) presents significant challenges, as models grow and evolve with complex inheritance relationships. This complexity is amplified when blockchain is employed to ensure the integrity and traceability of FL, where the need to edit multiple interlinked blockchain records and update all inherited models complicates the process.In this paper, we introduce Blockchained Federated Unlearning (BlockFUL), a novel framework with a dual-chain structure comprising a live chain and an archive chain for enabling unlearning capabilities within Blockchained FL. BlockFUL introduces two new unlearning paradigms, i.e., parallel and sequential paradigms, which can be effectively implemented through gradient-ascent-based and re-training-based unlearning methods. These methods enhance the unlearning process across multiple inherited models by enabling efficient consensus operations and reducing computational costs. Our extensive experiments validate that these methods effectively reduce data dependency and operational overhead, thereby boosting the overall performance of unlearning inherited models within BlockFUL on CIFAR-10 and Fashion-MNIST datasets using AlexNet, ResNet18, and MobileNetV2 models.

cs.CR↗

arXiv2025

Photonics-Aware Planning-Guided Automated Electrical Routing for Large-Scale Active Photonic Integrated Circuits

The rising demand for AI training and inference, as well as scientific computing, combined with stringent latency and energy budgets, is driving the adoption of integrated photonics for computing, sensing, and communications. As active photonic integrated circuits (PICs) scale in device count and functional heterogeneity, physical implementation by manual scripting and ad-hoc edits is no longer tenable. This creates an immediate need for an electronic-photonic design automation (EPDA) stack in which physical design automation is a core capability. However, there is currently no end-to-end fully automated routing flow that coordinates photonic waveguides and on-chip metal interconnect. Critically, available digital VLSI and analog/custom routers are not directly applicable to PIC metal routing due to a lack of customization to handle constraints induced by photonic devices and waveguides. We present, to our knowledge, the first end-to-end routing framework for large-scale active PICs that jointly addresses waveguides and metal wires within a unified flow. We introduce a physically-aware global planner that generates congestion- and crossing-aware routing guides while explicitly accounting for the placement of photonic components and waveguides. We further propose a sequence-consistent track assignment and a soft guidance-assisted detailed routing to speed up the routing process with significantly optimized routability and via usage. Evaluated on various large PIC designs, our router delivers fast, high-quality active PIC routing solutions with fewer vias, lower congestion, and competitive runtime relative to manual and existing VLSI router baselines; on average it reduce via count by ~99%, user-specified design rule violation by ~98%, and runtime by 17x, establishing a practical foundation for EPDA at system scale.

physics.optics↗

arXiv2025

Uniqueness of inverse random source problems for stochastic heat and wave equations

This paper investigates an inverse random source problem for stochastic evolution equations, including stochastic heat and wave equations, with the unknown source modeled as $g(x)f(t)\dot{W}(t)$. The research commences with the establishment of the well-posedness of the corresponding stochastic direct problem. Under suitable regularity conditions, the existence of stochastic strong solutions for both the stochastic heat and wave equations is demonstrated. For the inverse problem, the objective is to uniquely recover the strength $|f(t)|$ of the time-dependent component of the source from the boundary flux on a nonempty open subset. The uniqueness of the recovery for both the stochastic heat and wave equations is proven, and several numerical examples are given to verify the theoretical results.

math.AP↗

arXiv2025

BaseReward: A Strong Baseline for Multimodal Reward Model

The rapid advancement of Multimodal Large Language Models (MLLMs) has made aligning them with human preferences a critical challenge. Reward Models (RMs) are a core technology for achieving this goal, but a systematic guide for building state-of-the-art Multimodal Reward Models (MRMs) is currently lacking in both academia and industry. Through exhaustive experimental analysis, this paper aims to provide a clear ``recipe'' for constructing high-performance MRMs. We systematically investigate every crucial component in the MRM development pipeline, including \textit{reward modeling paradigms} (e.g., Naive-RM, Critic-based RM, and Generative RM), \textit{reward head architecture}, \textit{training strategies}, \textit{data curation} (covering over ten multimodal and text-only preference datasets), \textit{backbone model} and \textit{model scale}, and \textit{ensemble methods}. Based on these experimental insights, we introduce \textbf{BaseReward}, a powerful and efficient baseline for multimodal reward modeling. BaseReward adopts a simple yet effective architecture, built upon a {Qwen2.5-VL} backbone, featuring an optimized two-layer reward head, and is trained on a carefully curated mixture of high-quality multimodal and text-only preference data. Our results show that BaseReward establishes a new SOTA on major benchmarks such as MM-RLHF-Reward Bench, VL-Reward Bench, and Multimodal Reward Bench, outperforming previous models. Furthermore, to validate its practical utility beyond static benchmarks, we integrate BaseReward into a real-world reinforcement learning pipeline, successfully enhancing an MLLM's performance across various perception, reasoning, and conversational tasks. This work not only delivers a top-tier MRM but, more importantly, provides the community with a clear, empirically-backed guide for developing robust reward models for the next generation of MLLMs.

cs.CV↗

arXiv2025

A primal-dual price-optimization method for computing equilibrium prices in mean-field games models

We develop a simple yet efficient Lagrangian method for computing equilibrium prices in a mean-field game price-formation model. We prove that equilibrium prices are optimal in terms of a suitable criterion and derive a primal-dual gradient-based algorithm for computing them. One of the highlights of our computational framework is the efficient, simple, and flexible implementation of the algorithm using modern automatic differentiation techniques. Our implementation is modular and admits a seamless extension to high-dimensional settings with more complex dynamics, costs, and equilibrium conditions. Additionally, automatic differentiation enables a versatile algorithm that requires only coding the cost functions of agents. It automatically handles the gradients of the costs, thereby eliminating the need to manually form the adjoint equations.

math.OC↗

arXiv2025

More AI Assistance Reduces Cognitive Engagement: Examining the AI Assistance Dilemma in AI-Supported Note-Taking

As AI tools become increasingly embedded in cognitively demanding tasks such as note-taking, questions remain about whether they enhance or undermine cognitive engagement. This paper examines the "AI Assistance Dilemma" in note-taking, investigating how varying levels of AI support affect user engagement and comprehension. In a within-subject experiment, we asked participants (N=30) to take notes during lecture videos under three conditions: Automated AI (high assistance with structured notes), Intermediate AI (moderate assistance with real-time summary, and Minimal AI (low assistance with transcript). Results reveal that Intermediate AI yields the highest post-test scores and Automated AI the lowest. Participants, however, preferred the automated setup due to its perceived ease of use and lower cognitive effort, suggesting a discrepancy between preferred convenience and cognitive benefits. Our study provides insights into designing AI assistance that preserves cognitive engagement, offering implications for designing moderate AI support in cognitive tasks.

cs.HC↗

arXiv2025

Tech-ASan: Two-stage check for Address Sanitizer

Address Sanitizer (ASan) is a sharp weapon for detecting memory safety violations, including temporal and spatial errors hidden in C/C++ programs during execution. However, ASan incurs significant runtime overhead, which limits its efficiency in testing large software. The overhead mainly comes from sanitizer checks due to the frequent and expensive shadow memory access. Over the past decade, many methods have been developed to speed up ASan by eliminating and accelerating sanitizer checks, however, they either fail to adequately eliminate redundant checks or compromise detection capabilities. To address this issue, this paper presents Tech-ASan, a two-stage check based technique to accelerate ASan with safety assurance. First, we propose a novel two-stage check algorithm for ASan, which leverages magic value comparison to reduce most of the costly shadow memory accesses. Second, we design an efficient optimizer to eliminate redundant checks, which integrates a novel algorithm for removing checks in loops. Third, we implement Tech-ASan as a memory safety tool based on the LLVM compiler infrastructure. Our evaluation using the SPEC CPU2006 benchmark shows that Tech-ASan outperforms the state-of-the-art methods with 33.70% and 17.89% less runtime overhead than ASan and ASan--, respectively. Moreover, Tech-ASan detects 56 fewer false negative cases than ASan and ASan-- when testing on the Juliet Test Suite under the same redzone setting.

cs.SE↗

arXiv2025

Weakly-Supervised 3D Visual Grounding based on Visual Language Alignment

Learning to ground natural language queries to target objects or regions in 3D point clouds is quite essential for 3D scene understanding. Nevertheless, existing 3D visual grounding approaches require a substantial number of bounding box annotations for text queries, which is time-consuming and labor-intensive to obtain. In this paper, we propose 3D-VLA, a weakly supervised approach for 3D visual grounding based on Visual Linguistic Alignment. Our 3D-VLA exploits the superior ability of current large-scale vision-language models (VLMs) on aligning the semantics between texts and 2D images, as well as the naturally existing correspondences between 2D images and 3D point clouds, and thus implicitly constructs correspondences between texts and 3D point clouds with no need for fine-grained box annotations in the training procedure. During the inference stage, the learned text-3D correspondence will help us ground the text queries to the 3D target objects even without 2D images. To the best of our knowledge, this is the first work to investigate 3D visual grounding in a weakly supervised manner by involving large scale vision-language models, and extensive experiments on ReferIt3D and ScanRefer datasets demonstrate that our 3D-VLA achieves comparable and even superior results over the fully supervised methods.

cs.CV↗

arXiv2025

Camera Pose Refinement via 3D Gaussian Splatting

Camera pose refinement aims at improving the accuracy of initial pose estimation for applications in 3D computer vision. Most refinement approaches rely on 2D-3D correspondences with specific descriptors or dedicated networks, requiring reconstructing the scene again for a different descriptor or fully retraining the network for each scene. Some recent methods instead infer pose from feature similarity, but their lack of geometry constraints results in less accuracy. To overcome these limitations, we propose a novel camera pose refinement framework leveraging 3D Gaussian Splatting (3DGS), referred to as GS-SMC. Given the widespread usage of 3DGS, our method can employ an existing 3DGS model to render novel views, providing a lightweight solution that can be directly applied to diverse scenes without additional training or fine-tuning. Specifically, we introduce an iterative optimization approach, which refines the camera pose using epipolar geometric constraints among the query and multiple rendered images. Our method allows flexibly choosing feature extractors and matchers to establish these constraints. Extensive empirical evaluations on the 7-Scenes and the Cambridge Landmarks datasets demonstrate that our method outperforms state-of-the-art camera pose refinement approaches, achieving 53.3% and 56.9% reductions in median translation and rotation errors on 7-Scenes, and 40.7% and 53.2% on Cambridge.

cs.CV↗

arXiv2025

ZAPS: A Zero-Knowledge Proof Protocol for Secure UAV Authentication with Flight Path Privacy

The increasing deployment of Unmanned Aerial Vehicles (UAVs) for military, commercial, and logistics applications has raised significant concerns regarding flight path privacy. Conventional UAV communication systems often expose flight path data to third parties, making them vulnerable to tracking, surveillance, and location inference attacks. Existing encryption techniques provide security but fail to ensure complete privacy, as adversaries can still infer movement patterns through metadata analysis. To address these challenges, we propose a zk-SNARK(Zero-Knowledge Succinct Non-Interactive Argument of Knowledge)-based privacy-preserving flight path authentication and verification framework. Our approach ensures that a UAV can prove its authorisation, validate its flight path with a control centre, and comply with regulatory constraints without revealing any sensitive trajectory information. By leveraging zk-SNARKs, the UAV can generate cryptographic proofs that verify compliance with predefined flight policies while keeping the exact path and location undisclosed. This method mitigates risks associated with real-time tracking, identity exposure, and unauthorised interception, thereby enhancing UAV operational security in adversarial environments. Our proposed solution balances privacy, security, and computational efficiency, making it suitable for resource-constrained UAVs in both civilian and military applications.

cs.CR↗

arXiv2025

Finite Sample Analysis of System Poles for Ho-Kalman Algorithm

The Ho-Kalman algorithm has been widely employed for the identification of discrete-time linear time-invariant (LTI) systems. In this paper, we investigate the pole estimation error for the Ho-Kalman algorithm based on finite input/output sample data. Building upon prior works, we derive finite sample error bounds for system pole estimation in both single-trajectory and multiple-trajectory scenarios. Specifically, we prove that, with high probability, the estimation error for an $n$-dimensional system decreases at a rate of at least $\mathcal{O}(T^{-1/2n})$ in the single-trajectory setting with trajectory length $T$, and at a rate of at least $\mathcal{O}(N^{-1/2n})$ in the multiple-trajectory setting with $N$ independent trajectories. Furthermore, we reveal that in both settings, achieving a constant estimation error requires a super-polynomial sample size in $ \max\{n/m, n/p\} $, where $n/m$ and $n/p$ denote the state-to-output and state-to-input dimension ratios, respectively. Finally, numerical experiments are conducted to validate the non-asymptotic results of system pole estimation.

eess.SY↗

arXiv2025

SLRTP2025 Sign Language Production Challenge: Methodology, Results, and Future Work

Sign Language Production (SLP) is the task of generating sign language video from spoken language inputs. The field has seen a range of innovations over the last few years, with the introduction of deep learning-based approaches providing significant improvements in the realism and naturalness of generated outputs. However, the lack of standardized evaluation metrics for SLP approaches hampers meaningful comparisons across different systems. To address this, we introduce the first Sign Language Production Challenge, held as part of the third SLRTP Workshop at CVPR 2025. The competition's aims are to evaluate architectures that translate from spoken language sentences to a sequence of skeleton poses, known as Text-to-Pose (T2P) translation, over a range of metrics. For our evaluation data, we use the RWTH-PHOENIX-Weather-2014T dataset, a German Sign Language - Deutsche Gebardensprache (DGS) weather broadcast dataset. In addition, we curate a custom hidden test set from a similar domain of discourse. This paper presents the challenge design and the winning methodologies. The challenge attracted 33 participants who submitted 231 solutions, with the top-performing team achieving BLEU-1 scores of 31.40 and DTW-MJE of 0.0574. The winning approach utilized a retrieval-based framework and a pre-trained language model. As part of the workshop, we release a standardized evaluation network, including high-quality skeleton extraction-based keypoints establishing a consistent baseline for the SLP field, which will enable future researchers to compare their work against a broader range of methods.

cs.CV↗