arXiv subjects

Yuwei Li

Publications and source records attributed to Yuwei Li.

At least 19 recordsLinked to original sources

arXiv2026

EATR-Stereo: Embodiment-Aware Token Routing of Paired Stereo Evidence for Humanoid Vision-Language-Action Control

Long-horizon humanoid vision--language--action (VLA) control with head-mounted stereo cameras requires visual interfaces that can exploit complementary views while maintaining compatibility with pretrained representations. Existing interfaces often discard complementary stereo evidence or fuse additional observations without preserving the native primary-view pathway and adapting auxiliary information to robot embodiment. We present EATR-Stereo, an embodiment-aware token-routing framework that retains primary-view tokens and constructs primary-aligned Cross-View Auxiliary Tokens (CVATs) by querying the synchronized auxiliary-view token sequence. A body-segmented proprioceptive encoder further conditions token-wise auxiliary usage on robot configuration history, enabling selective incorporation of stereo evidence during action generation. The routed auxiliary stream augments the language and primary-visual context of a pretrained VLA while keeping its vision--language model frozen. On a 33-DoF physical humanoid with a 37-D proprioceptive state, we evaluate nine configurations in over-100-s search--approach--grasp--place--return tasks. EATR-Stereo achieves 60.0% full-task success, 100.0% grasp success, and 80.0% stage success. Under severe asymmetric occlusion, it improves recovery to 80% compared with 30% for CVAT alone. Ablation studies further show the importance of preserving primary tokens and combining cross-view auxiliary features with structured proprioceptive routing. These results demonstrate that selectively routed paired stereo evidence improves spatial grounding for reliable long-horizon humanoid VLA control.

cs.RO↗

arXiv2026

LPM 1.0: Video-based Character Performance Model

Performance, the externalization of intent, emotion, and personality through visual, vocal, and temporal behavior, is what makes a character alive. Learning such performance from video is a promising alternative to traditional 3D pipelines. However, existing video models struggle to jointly achieve high expressiveness, real-time inference, and long-horizon identity stability, a tension we call the performance trilemma. Conversation is the most comprehensive performance scenario, as characters simultaneously speak, listen, react, and emote while maintaining identity over time. To address this, we present LPM 1.0 (Large Performance Model), focusing on single-person full-duplex audio-visual conversational performance. Concretely, we build a multimodal human-centric dataset through strict filtering, speaking-listening audio-video pairing, performance understanding, and identity-aware multi-reference extraction; train a 17B-parameter Diffusion Transformer (Base LPM) for highly controllable, identity-consistent performance through multimodal conditioning; and distill it into a causal streaming generator (Online LPM) for low-latency, infinite-length interaction. At inference, given a character image with identity-aware references, LPM 1.0 generates listening videos from user audio and speaking videos from synthesized audio, with text prompts for motion control, all at real-time speed with identity-stable, infinite-length generation. LPM 1.0 thus serves as a visual engine for conversational agents, live streaming characters, and game NPCs. To systematically evaluate this setting, we propose LPM-Bench, the first benchmark for interactive character performance. LPM 1.0 achieves state-of-the-art results across all evaluated dimensions while maintaining real-time inference.

cs.CV↗

arXiv2025

Diffusion-Based Restoration for Multi-Modal 3D Object Detection in Adverse Weather

Multi-modal 3D object detection is important for reliable perception in robotics and autonomous driving. However, its effectiveness remains limited under adverse weather conditions due to weather-induced distortions and misalignment between different data modalities. In this work, we propose DiffFusion, a novel framework designed to enhance robustness in challenging weather through diffusion-based restoration and adaptive cross-modal fusion. Our key insight is that diffusion models possess strong capabilities for denoising and generating data that can adapt to various weather conditions. Building on this, DiffFusion introduces Diffusion-IR restoring images degraded by weather effects and Point Cloud Restoration (PCR) compensating for corrupted LiDAR data using image object cues. To tackle misalignments between two modalities, we develop Bidirectional Adaptive Fusion and Alignment Module (BAFAM). It enables dynamic multi-modal fusion and bidirectional bird's-eye view (BEV) alignment to maintain consistent spatial correspondence. Extensive experiments on three public datasets show that DiffFusion achieves state-of-the-art robustness under adverse weather while preserving strong clean-data performance. Zero-shot results on the real-world DENSE dataset further validate its generalization. The implementation of our DiffFusion will be released as open-source.

cs.CV↗

arXiv2025

ZSPAPrune: Zero-Shot Prompt-Aware Token Pruning for Vision-Language Models

As the capabilities of Vision-Language Models (VLMs) advance, they can process increasingly large inputs, which, unlike in LLMs, generates significant visual token redundancy and leads to prohibitive inference costs. While many methods aim to reduce these costs by pruning visual tokens, existing approaches, whether based on attention or diversity, typically neglect the guidance of the text prompt and thus fail to prioritize task relevance. In this work, we propose a novel, zero-shot method that reframes the problem by introducing a prompt-aware perspective, explicitly modeling visual token pruning as a balance between task relevance and information diversity. Our hierarchical approach first selects a core set of task-relevant visual tokens and then supplements them with diversity tokens to preserve broader context. Experiments across multiple models and benchmarks show that our method achieves performance that matches or surpasses the state-of-the-art with only minimal accuracy loss, even when pruning up to 90\% of the tokens. Furthermore, these gains are accompanied by significant reductions in GPU memory footprint and inference latency.

cs.CV↗

arXiv2025

MirrorFuzz: Leveraging LLM and Shared Bugs for Deep Learning Framework APIs Fuzzing

Deep learning (DL) frameworks serve as the backbone for a wide range of artificial intelligence applications. However, bugs within DL frameworks can cascade into critical issues in higher-level applications, jeopardizing reliability and security. While numerous techniques have been proposed to detect bugs in DL frameworks, research exploring common API patterns across frameworks and the potential risks they entail remains limited. Notably, many DL frameworks expose similar APIs with overlapping input parameters and functionalities, rendering them vulnerable to shared bugs, where a flaw in one API may extend to analogous APIs in other frameworks. To address this challenge, we propose MirrorFuzz, an automated API fuzzing solution to discover shared bugs in DL frameworks. MirrorFuzz operates in three stages: First, MirrorFuzz collects historical bug data for each API within a DL framework to identify potentially buggy APIs. Second, it matches each buggy API in a specific framework with similar APIs within and across other DL frameworks. Third, it employs large language models (LLMs) to synthesize code for the API under test, leveraging the historical bug data of similar APIs to trigger analogous bugs across APIs. We implement MirrorFuzz and evaluate it on four popular DL frameworks (TensorFlow, PyTorch, OneFlow, and Jittor). Extensive evaluation demonstrates that MirrorFuzz improves code coverage by 39.92\% and 98.20\% compared to state-of-the-art methods on TensorFlow and PyTorch, respectively. Moreover, MirrorFuzz discovers 315 bugs, 262 of which are newly found, and 80 bugs are fixed, with 52 of these bugs assigned CNVD IDs.

cs.SE↗

arXiv2025

GradEscape: A Gradient-Based Evader Against AI-Generated Text Detectors

In this paper, we introduce GradEscape, the first gradient-based evader designed to attack AI-generated text (AIGT) detectors. GradEscape overcomes the undifferentiable computation problem, caused by the discrete nature of text, by introducing a novel approach to construct weighted embeddings for the detector input. It then updates the evader model parameters using feedback from victim detectors, achieving high attack success with minimal text modification. To address the issue of tokenizer mismatch between the evader and the detector, we introduce a warm-started evader method, enabling GradEscape to adapt to detectors across any language model architecture. Moreover, we employ novel tokenizer inference and model extraction techniques, facilitating effective evasion even in query-only access. We evaluate GradEscape on four datasets and three widely-used language models, benchmarking it against four state-of-the-art AIGT evaders. Experimental results demonstrate that GradEscape outperforms existing evaders in various scenarios, including with an 11B paraphrase model, while utilizing only 139M parameters. We have successfully applied GradEscape to two real-world commercial AIGT detectors. Our analysis reveals that the primary vulnerability stems from disparity in text expression styles within the training data. We also propose a potential defense strategy to mitigate the threat of AIGT evaders. We open-source our GradEscape for developing more robust AIGT detectors.

cs.CR↗

arXiv2025

Identification of the high-pressure phases of alpha-SnWO4 combining x-ray diffraction and crystal structure prediction

We have characterized the high-pressure behavior of alpha-SnWO4. The compound has been studied up to 30 GPa using a diamond-anvil cell and synchrotron powder X-ray diffraction. We report evidence of two structural phase transitions in the pressure range covered in our study, and we propose a crystal structure for the two high-pressure phases. The first one, observed around 12.9 GPa, has been obtained combining indexation using DICVOL and density-functional theory calculations. The second high-pressure phase, observed around 17.5 GPa, has been determined by using the CALYPSO code, the prediction of which was supported by a Le Bail fit to the experimental X-ray diffraction patterns. The proposed structural sequence involves two successive collapses of the unit-cell volume and an increase in the coordination number of Sn and W atoms. The room-temperature equations of state, the principal axes of compression and their compressibility, the elastic constants, and the elastic moduli are reported for {\alpha}-SnWO4 and for the two high-pressure phases.

cond-mat.mtrl-sci↗

arXiv2025

Dependence of the intensity of the nonwave component of EUV waves on coronal magnetic field configuration

Context. Mounting evidence has shown that EUV waves consist of a fast-mode magnetohydrodynamic (MHD) wave (or shock wave) followed by a slower nonwave component, as predicted by the magnetic fieldline stretching model. However, not all observed events display both wavefronts, particularly the slower nonwave component. Even in case that the slower nonwave component is present, the intensity distribution often exhibits strong anisotropy. Aims. This study is intended to unveil the formation condition of the slower nonwave component of EUV waves. Methods. We analyzed the EUV wave event on 8 March 2019, and compared the EUV wave intensity map with the extrapolation coronal potential magnetic field. Data-inspired MHD simulation was also performed. Results. Two types of EUV waves are identified, and the slower nonwave component exhibits strong anisotropy. By reconstructing 3D coronal magnetic fields, we found that the slower nonwave component of EUV waves is more pronounced in the regions where magnetic fields are backward-inclined, which is further reproduced by our MHD simulations. Conclusions. The anisotropy of the slower nonwave component of EUV waves is strongly related to the magnetic configuration, with backward-inclined field lines favoring their appearance. The more the field lines are forward-inclined, the weaker such wavelike fronts are.

astro-ph.SR↗

arXiv2025

Study on impact mechanism and precursor information induced by high intensity mining

With heightened mining intensity, the incidence of coal bursts is escalating, necessitating advanced understanding and prediction techniques. This research delves into the intricacies of coal burst mechanisms, proposing a novel theoretical model for the release of coal mass energy founded on the tenets of stress superposition. A significant revelation is that the energy culminating in a coal burst is an amalgamation of intrinsic coal strain energy and perturbations from mining activities. Field investigations scrutinize the microseismic parameters across a spectrum of mining velocities, discerning potential failure regions and precursor hallmarks in high-intensity mining environments. Notably, microseismic energy, in such contexts, experiences an augmentation of approximately 2000 J. Numerical simulations executed via 3DEC elucidate stress distribution patterns and failure modalities of adjacent rock structures in relation to mining velocities. The simulations underscore that an uptick in mining speed diminishes the buffer to high-pressure abutments, intensifying inherent pressures. For mitigation, it's advocated that high-intensity mining advances be capped at 11 m/d. Merging theoretical analysis, experimental data, field assessments, and computational simulations, this study proffers a holistic insight into coal burst dynamics, underscoring its value in refining monitoring and early warning protocols in the domain.

physics.geo-ph↗

arXiv2025

Dialogue Injection Attack: Jailbreaking LLMs through Context Manipulation

Large language models (LLMs) have demonstrated significant utility in a wide range of applications; however, their deployment is plagued by security vulnerabilities, notably jailbreak attacks. These attacks manipulate LLMs to generate harmful or unethical content by crafting adversarial prompts. While much of the current research on jailbreak attacks has focused on single-turn interactions, it has largely overlooked the impact of historical dialogues on model behavior. In this paper, we introduce a novel jailbreak paradigm, Dialogue Injection Attack (DIA), which leverages the dialogue history to enhance the success rates of such attacks. DIA operates in a black-box setting, requiring only access to the chat API or knowledge of the LLM's chat template. We propose two methods for constructing adversarial historical dialogues: one adapts gray-box prefilling attacks, and the other exploits deferred responses. Our experiments show that DIA achieves state-of-the-art attack success rates on recent LLMs, including Llama-3.1 and GPT-4o. Additionally, we demonstrate that DIA can bypass 5 different defense mechanisms, highlighting its robustness and effectiveness.

cs.CL↗

arXiv2024

G-Fuzz: A Directed Fuzzing Framework for gVisor

gVisor is a Google-published application-level kernel for containers. As gVisor is lightweight and has sound isolation, it has been widely used in many IT enterprises \cite{Stripe, DigitalOcean, Cloundflare}. When a new vulnerability of the upstream gVisor is found, it is important for the downstream developers to test the corresponding code to maintain the security. To achieve this aim, directed fuzzing is promising. Nevertheless, there are many challenges in applying existing directed fuzzing methods for gVisor. The core reason is that existing directed fuzzers are mainly for general C/C++ applications, while gVisor is an OS kernel written in the Go language. To address the above challenges, we propose G-Fuzz, a directed fuzzing framework for gVisor. There are three core methods in G-Fuzz, including lightweight and fine-grained distance calculation, target related syscall inference and utilization, and exploration and exploitation dynamic switch. Note that the methods of G-Fuzz are general and can be transferred to other OS kernels. We conduct extensive experiments to evaluate the performance of G-Fuzz. Compared to Syzkaller, the state-of-the-art kernel fuzzer, G-Fuzz outperforms it significantly. Furthermore, we have rigorously evaluated the importance for each core method of G-Fuzz. G-Fuzz has been deployed in industry and has detected multiple serious vulnerabilities.

cs.CR↗

arXiv2024

An Empirical Study on the Distance Metric in Guiding Directed Grey-box Fuzzing

Directed grey-box fuzzing (DGF) aims to discover vulnerabilities in specific code areas efficiently. Distance metric, which is used to measure the quality of seed in DGF, is a crucial factor in affecting the fuzzing performance. Despite distance metrics being widely applied in existing DGF frameworks, it remains opaque about how different distance metrics guide the fuzzing process and affect the fuzzing result in practice. In this paper, we conduct the first empirical study to explore how different distance metrics perform in guiding DGFs. Specifically, we systematically discuss different distance metrics in the aspect of calculation method and granularity. Then, we implement different distance metrics based on AFLGo. On this basis, we conduct comprehensive experiments to evaluate the performance of these distance metrics on the benchmarks widely used in existing DGF-related work. The experimental results demonstrate the following insights. First, the difference among different distance metrics with varying methods of calculation and granularities is not significant. Second, the distance metrics may not be effective in describing the difficulty of triggering the target vulnerability. In addition, by scrutinizing the quality of testcases, our research highlights the inherent limitation of existing mutation strategies in generating high-quality testcases, calling for designing effective mutation strategies for directed fuzzing. We open-source the implementation code and experiment dataset to facilitate future research in DGF.

cs.CR↗

arXiv2023

The asymptotic distribution of maxima of stationary random sequences under random replacing

In this paper, we investigated the effect on extreme of random replacing for a stationary sequence satisfying a type of long dependent condition and a local dependent condition, and derived the joint asymptotic distribution of maximum from the stationary sequence and the maximum from the random replacing sequence. We also provided several applications for our main results.

math.PR↗

arXiv2023

A novel approach for bilevel programs based on Wolfe duality

This paper considers a bilevel program, which has many applications in practice. To develop effective numerical algorithms, it is generally necessary to transform the bilevel program into a single-level optimization problem. The most popular approach is to replace the lower-level program by its KKT conditions and then the bilevel program can be reformulated as a mathematical program with equilibrium constraints (MPEC for short). However, since the MPEC does not satisfy the Mangasarian-Fromovitz constraint qualification at any feasible point, the well-developed nonlinear programming theory cannot be applied to MPECs directly. In this paper, we apply the Wolfe duality to show that, under very mild conditions, the bilevel program is equivalent to a new single-level reformulation (WDP for short) in the globally and locally optimal sense. We give an example to show that, unlike the MPEC reformulation, WDP may satisfy the Mangasarian-Fromovitz constraint qualification at its feasible points. We give some properties of the WDP reformulation and the relations between the WDP and MPEC reformulations. We further propose a relaxation method for solving WDP and investigate its limiting behavior. Comprehensive numerical experiments indicate that, although solving WDP directly does not perform very well in our tests, the relaxation method based on the WDP reformulation is quite efficient.

math.OC↗

arXiv2023

Learning Trustworthy Model from Noisy Labels based on Rough Set for Surface Defect Detection

In the surface defect detection, there are some suspicious regions that cannot be uniquely classified as abnormal or normal. The annotating of suspicious regions is easily affected by factors such as workers' emotional fluctuations and judgment standard, resulting in noisy labels, which in turn leads to missing and false detections, and ultimately leads to inconsistent judgments of product quality. Unlike the usual noisy labels, the ones used for surface defect detection appear to be inconsistent rather than mislabeled. The noise occurs in almost every label and is difficult to correct or evaluate. In this paper, we proposed a framework that learns trustworthy models from noisy labels for surface defect defection. At first, to avoid the negative impact of noisy labels on the model, we represent the suspicious regions with consistent and precise elements at the pixel-level and redesign the loss function. Secondly, without changing network structure and adding any extra labels, pluggable spatially correlated Bayesian module is proposed. Finally, the defect discrimination confidence is proposed to measure the uncertainty, with which anomalies can be identified as defects. Our results indicate not only the effectiveness of the proposed method in learning from noisy labels, but also robustness and real-time performance.

cs.CV↗

arXiv2022

SCULPTOR: Skeleton-Consistent Face Creation Using a Learned Parametric Generator

Recent years have seen growing interest in 3D human faces modelling due to its wide applications in digital human, character generation and animation. Existing approaches overwhelmingly emphasized on modeling the exterior shapes, textures and skin properties of faces, ignoring the inherent correlation between inner skeletal structures and appearance. In this paper, we present SCULPTOR, 3D face creations with Skeleton Consistency Using a Learned Parametric facial generaTOR, aiming to facilitate easy creation of both anatomically correct and visually convincing face models via a hybrid parametric-physical representation. At the core of SCULPTOR is LUCY, the first large-scale shape-skeleton face dataset in collaboration with plastic surgeons. Named after the fossils of one of the oldest known human ancestors, our LUCY dataset contains high-quality Computed Tomography (CT) scans of the complete human head before and after orthognathic surgeries, critical for evaluating surgery results. LUCY consists of 144 scans of 72 subjects (31 male and 41 female) where each subject has two CT scans taken pre- and post-orthognathic operations. Based on our LUCY dataset, we learn a novel skeleton consistent parametric facial generator, SCULPTOR, which can create the unique and nuanced facial features that help define a character and at the same time maintain physiological soundness. Our SCULPTOR jointly models the skull, face geometry and face appearance under a unified data-driven framework, by separating the depiction of a 3D face into shape blend shape, pose blend shape and facial expression blend shape. SCULPTOR preserves both anatomic correctness and visual realism in facial generation tasks compared with existing methods. Finally, we showcase the robustness and effectiveness of SCULPTOR in various fancy applications unseen before.

cs.CV↗

arXiv2022

Joint Offloading Decision and Resource Allocation for Vehicular Fog-Edge Computing Networks: A Contract-Stackelberg Approach

With the popularity of mobile devices and development of computationally intensive applications, researchers are focusing on offloading computation to Mobile Edge Computing (MEC) server due to its high computational efficiency and low communication delay. As the computing resources of an MEC server are limited, vehicles in the urban area who have abundant idle resources should be fully utilized. However, offloading computing tasks to vehicles faces many challenging issues. In this paper, we introduce a vehicular fog-edge computing paradigm and formulate it as a multi-stage Stackelberg game to deal with these issues. Specifically, vehicles are not obligated to share resources, let alone disclose their private information (e.g., stay time and the amount of resources). Therefore, in the first stage, we design a contract-based incentive mechanism to motivate vehicles to contribute their idle resources. Next, due to the complicated interactions among vehicles, road-side unit (RSU), MEC server and mobile device users, it is challenging to coordinate the resources of all parties and design a transaction mechanism to make all entities benefit. In the second and third stages, based on Stackelberg game, we develop pricing strategies that maximize the utilities of all parties. The analytical forms of optimal strategies for each stage are given. Simulation results demonstrate the effectiveness of our proposed incentive mechanism, reveal the trends of energy consumption and offloading decisions of users with various parameters, and present the performance comparison between our framework and existing MEC offloading paradigm in vehicular networks.

cs.DC↗

arXiv2022

NIMBLE: A Non-rigid Hand Model with Bones and Muscles

Emerging Metaverse applications demand reliable, accurate, and photorealistic reproductions of human hands to perform sophisticated operations as if in the physical world. While real human hand represents one of the most intricate coordination between bones, muscle, tendon, and skin, state-of-the-art techniques unanimously focus on modeling only the skeleton of the hand. In this paper, we present NIMBLE, a novel parametric hand model that includes the missing key components, bringing 3D hand model to a new level of realism. We first annotate muscles, bones and skins on the recent Magnetic Resonance Imaging hand (MRI-Hand) dataset and then register a volumetric template hand onto individual poses and subjects within the dataset. NIMBLE consists of 20 bones as triangular meshes, 7 muscle groups as tetrahedral meshes, and a skin mesh. Via iterative shape registration and parameter learning, it further produces shape blend shapes, pose blend shapes, and a joint regressor. We demonstrate applying NIMBLE to modeling, rendering, and visual inference tasks. By enforcing the inner bones and muscles to match anatomic and kinematic rules, NIMBLE can animate 3D hands to new poses at unprecedented realism. To model the appearance of skin, we further construct a photometric HandStage to acquire high-quality textures and normal maps to model wrinkles and palm print. Finally, NIMBLE also benefits learning-based hand pose and shape estimation by either synthesizing rich data or acting directly as a differentiable layer in the inference network.

cs.CV↗