arXiv subjects

Zheyuan Yang

Publications and source records attributed to Zheyuan Yang.

14 recordsLinked to original sources

arXiv2026

GUI-CC: Benchmarking Contextual Consistency of GUI World Models as Agent Environments

GUI world models are increasingly evaluated as one-step next-screen predictors, yet their intended use is often as multi-step environments for GUI agents. This mismatch leaves a key requirement under-tested: generated states must remain contextually consistent when they are repeatedly reused for future interaction. We introduce GUI-CC, a benchmark that evaluates contextual consistency of GUI world models as agent environments rather than isolated next-screen predictors. GUI-CC contains two complementary tracks: an offline reference-action track that rolls models along real mobile GUI trajectories, and an online agent-loop track that lets fixed probing agents interact with model-generated UIs. We construct 500 offline trajectory tasks from GUIOdyssey and 200 emulator-verified online tasks across 30 mobile apps. GUI-CC evaluates transition fidelity, transition plausibility, contextual consistency, and task progress. Experiments show that plausible single-step generation does not guarantee reliable environment simulation: current models often produce usable-looking screens while failing to preserve task-relevant context or support executable multi-step rollouts.

cs.CL↗

arXiv2026

Are Android GUI Agents Robust Against Runtime Anomalies? AnTrap: Evaluating Agents in Dynamic Adversarial Environments

GUI agents often encounter dynamic anomalies when deployed on Android devices, from unexpected pop-ups to action misuse, yet existing benchmarks lack systematic evaluation of agent robustness against runtime anomalies. We introduce AnTrap, a comprehensive benchmark that injects dynamic perturbations into agent execution trajectories. We propose a taxonomy organizing real-world anomalies into four layers (State, Thinking, Action and Round) with ten fine-grained subcategories, and develop a construction pipeline that preserves task solvability while introducing realistic adversarial conditions. Evaluating 16 leading GUI models, we reveal universal vulnerability to dynamic anomalies, with even the strongest models suffering significant performance degradation. Furthermore, we conduct GRPO training in both original and adversarial environments to validate our benchmark, separating environment-learnable anomalies from reasoning-bottlenecked ones. Our findings show that while single-step traps at state and action layers are largely addressable through adversarial reinforcement learning, deep contextual traps, like state deadlock, expose intrinsic limitations that cannot be resolved by training in environments with traps alone.

cs.AI↗

arXiv2026

Sci-VBench: Evaluating Knowledge- and Reasoning-Intensive Video Generation in Science Domains

We introduce Sci-VBench, a comprehensive benchmark for evaluating knowledge- and reasoning-intensive video generation across scientific domains. It contains 1,253 expert-annotated examples spanning 60 subjects across four core disciplines: Natural Science, Healthcare, Humanities & Social Sciences, and Engineering. Each example requires models to generate temporally rich videos that demand scientific reasoning and knowledge-grounded synthesis, going beyond surface-level visual plausibility. We further establish a rubric-based evaluation protocol. Our analysis shows that, under this protocol, both non-expert human evaluators and MLLM-as-Judge systems can achieve relatively high agreement with expert judgments, supporting reproducible evaluation at scale. We benchmark 16 frontier proprietary and open-source models and find that, while automatic perceptual-quality scores cluster tightly across systems, performance on Prompt Grounding and Scientific and Causal Correctness varies substantially, with a pronounced proprietary-open-source gap. These findings show that advances in visual realism have not yet translated into reliable modeling of scientific and causal dynamics.

cs.CV↗

arXiv2026

VideoKR: Towards Knowledge- and Reasoning-Intensive Video Understanding

We introduce VideoKR, the first large-scale training corpus specifically designed to strengthen knowledge- and reasoning-intensive video understanding. It comprises 315K video reasoning examples over 145K newly collected, CC-licensed, expert-domain videos. We develop a human-in-the-loop, skill-oriented example generation pipeline that targets progressively deeper video reasoning capabilities while ensuring the difficulty, diversity, and reliability of both the examples and their CoT rationales. We also curate VideoKR-Eval, a new expert-annotated benchmark where questions require genuine video understanding and knowledge-intensive reasoning rather than textual shortcuts. Our experiments show that, under a standard SFT$\rightarrow$GRPO pipeline, models post-trained on VideoKR outperform prior post-training approaches on knowledge-intensive video reasoning while remaining competitive on general video reasoning, highlighting data design as a key driver of progress in video reasoning. We further conduct comprehensive ablations to isolate the contributions of VideoKR, providing actionable insights for future work.

cs.CV↗

arXiv2026

Parameter Efficient Multi-Class Intelligent Scheduling for Multimodal Online Distributed Industrial Anomaly Detection

Industrial anomaly detection has attracted significant attention as a fundamental challenge in industrial systems. The rapid advancement of heterogeneous industrial sensors has driven industrial anomaly detection from unimodal to multimodal paradigms. However, existing methods are primarily designed for centralized and offline settings, overlooking the distributed and continuously generated data characteristic of real-world industrial environments. With the advancement of edge intelligence, modern edge devices are increasingly capable of not only data acquisition but also distributed model training, enabling collaborative intelligence across the system. Industrial anomaly detection represents a critical application in this context. Motivated by these challenges, we propose a novel framework termed Multimodal Online Distributed Industrial Anomaly Detection (MODIAD). We first present a comprehensive workflow for MODIAD and then formulate a Multi-class Intelligent Scheduling (MIS) problem to coordinate cross class model updates by balancing data sufficiency and class update frequency. To efficiently solve this problem, we design a Sequential Marginal Gain Greedy (SMG) algorithm that enables effective multi-class training under resource constraints. Furthermore, to improve the computational and communication efficiency during training, we propose an Resource Efficient Class-Wise Low Rank Adaptation (REC-LoRA) strategy, which significantly reduces system overhead while preserving detection performance. Extensive experiments on two representative multimodal industrial anomaly detection datasets, MVTec 3D-AD and Eyecandies demonstrate that the proposed approach achieves superior performance and efficiency under the MODIAD scenario.

cs.LG↗

arXiv2026

TableVista: Benchmarking Multimodal Table Reasoning under Visual and Structural Complexity

We introduce TableVista, a comprehensive benchmark for evaluating foundation models in multimodal table reasoning under visual and structural complexity. TableVista consists of 3,000 high-quality table reasoning problems, where each instance is expanded into 10 distinct visual variants through our multi-style rendering and transformation pipeline. This process encompasses diverse scenario styles, robustness perturbations, and vision-only configurations, culminating in 30,000 multimodal samples for a multi-dimensional evaluation. We conduct an extensive evaluation of 29 state-of-the-art open-source and proprietary foundation models on TableVista. Through comprehensive quantitative and qualitative analysis, we find that while evaluated models remain largely stable across diverse rendering styles, they exhibit pronounced performance degradation on complex structural layouts and vision-only settings, revealing that current models struggle to maintain reasoning consistency when structural complexity combines with visually integrated presentations. These findings highlight critical gaps in current multimodal capabilities, providing insights for advancing more robust and reliable table understanding models.

cs.CL↗

arXiv2025

Can LLMs Generate High-Quality Test Cases for Algorithm Problems? TestCase-Eval: A Systematic Evaluation of Fault Coverage and Exposure

We introduce TestCase-Eval, a new benchmark for systematic evaluation of LLMs in test-case generation. TestCase-Eval includes 500 algorithm problems and 100,000 human-crafted solutions from the Codeforces platform. It focuses on two pivotal tasks: (1) Fault Coverage, which measures how well LLM-generated test sets probe diverse input scenarios and cover a wide range of potential failure modes. (2) Fault Exposure, which evaluates whether LLMs can craft a tailored test input that reveals a specific incorrect code implementation. We provide a comprehensive assessment of 19 state-of-the-art open-source and proprietary LLMs on TestCase-Eval, offering insights into their strengths and limitations in generating effective test cases for algorithm problems.

cs.SE↗

arXiv2025

Table-R1: Inference-Time Scaling for Table Reasoning

In this work, we present the first study to explore inference-time scaling on table reasoning tasks. We develop and evaluate two post-training strategies to enable inference-time scaling: distillation from frontier model reasoning traces and reinforcement learning with verifiable rewards (RLVR). For distillation, we introduce a large-scale dataset of reasoning traces generated by DeepSeek-R1, which we use to fine-tune LLMs into the Table-R1-SFT model. For RLVR, we propose task-specific verifiable reward functions and apply the GRPO algorithm to obtain the Table-R1-Zero model. We evaluate our Table-R1-series models across diverse table reasoning tasks, including short-form QA, fact verification, and free-form QA. Notably, the Table-R1-Zero model matches or exceeds the performance of GPT-4.1 and DeepSeek-R1, while using only a 7B-parameter LLM. It also demonstrates strong generalization to out-of-domain datasets. Extensive ablation and qualitative analyses reveal the benefits of instruction tuning, model architecture choices, and cross-task generalization, as well as emergence of essential table reasoning skills during RL training.

cs.CL↗

arXiv2024

UniGaussian: Driving Scene Reconstruction from Multiple Camera Models via Unified Gaussian Representations

Urban scene reconstruction is crucial for real-world autonomous driving simulators. Although existing methods have achieved photorealistic reconstruction, they mostly focus on pinhole cameras and neglect fisheye cameras. In fact, how to effectively simulate fisheye cameras in driving scene remains an unsolved problem. In this work, we propose UniGaussian, a novel approach that learns a unified 3D Gaussian representation from multiple camera models for urban scene reconstruction in autonomous driving. Our contributions are two-fold. First, we propose a new differentiable rendering method that distorts 3D Gaussians using a series of affine transformations tailored to fisheye camera models. This addresses the compatibility issue of 3D Gaussian splatting with fisheye cameras, which is hindered by light ray distortion caused by lenses or mirrors. Besides, our method maintains real-time rendering while ensuring differentiability. Second, built on the differentiable rendering method, we design a new framework that learns a unified Gaussian representation from multiple camera models. By applying affine transformations to adapt different camera models and regularizing the shared Gaussians with supervision from different modalities, our framework learns a unified 3D Gaussian representation with input data from multiple sources and achieves holistic driving scene understanding. As a result, our approach models multiple sensors (pinhole and fisheye cameras) and modalities (depth, semantic, normal and LiDAR point clouds). Our experiments show that our method achieves superior rendering quality and fast rendering speed for driving scene simulation.

cs.CV↗

arXiv2024

VQA-Diff: Exploiting VQA and Diffusion for Zero-Shot Image-to-3D Vehicle Asset Generation in Autonomous Driving

Generating 3D vehicle assets from in-the-wild observations is crucial to autonomous driving. Existing image-to-3D methods cannot well address this problem because they learn generation merely from image RGB information without a deeper understanding of in-the-wild vehicles (such as car models, manufacturers, etc.). This leads to their poor zero-shot prediction capability to handle real-world observations with occlusion or tricky viewing angles. To solve this problem, in this work, we propose VQA-Diff, a novel framework that leverages in-the-wild vehicle images to create photorealistic 3D vehicle assets for autonomous driving. VQA-Diff exploits the real-world knowledge inherited from the Large Language Model in the Visual Question Answering (VQA) model for robust zero-shot prediction and the rich image prior knowledge in the Diffusion model for structure and appearance generation. In particular, we utilize a multi-expert Diffusion Models strategy to generate the structure information and employ a subject-driven structure-controlled generation mechanism to model appearance information. As a result, without the necessity to learn from a large-scale image-to-3D vehicle dataset collected from the real world, VQA-Diff still has a robust zero-shot image-to-novel-view generation ability. We conduct experiments on various datasets, including Pascal 3D+, Waymo, and Objaverse, to demonstrate that VQA-Diff outperforms existing state-of-the-art methods both qualitatively and quantitatively.

cs.CV↗

arXiv2023

Learning Effective NeRFs and SDFs Representations with 3D Generative Adversarial Networks for 3D Object Generation

We present a solution for 3D object generation of ICCV 2023 OmniObject3D Challenge. In recent years, 3D object generation has made great process and achieved promising results, but it remains a challenging task due to the difficulty of generating complex, textured, and high-fidelity results. To resolve this problem, we study learning effective NeRFs and SDFs representations with 3D Generative Adversarial Networks (GANs) for 3D object generation. Specifically, inspired by recent works, we use the efficient geometry-aware 3D GANs as the backbone incorporating with label embedding and color mapping, which enables to train the model on different taxonomies simultaneously. Then, through a decoder, we aggregate the resulting features to generate Neural Radiance Fields (NeRFs) based representations for rendering high-fidelity synthetic images. Meanwhile, we optimize Signed Distance Functions (SDFs) to effectively represent objects with 3D meshes. Besides, we observe that this model can be effectively trained with only a few images of each object from a variety of classes, instead of using a great number of images per object or training one model per class. With this pipeline, we can optimize an effective model for 3D object generation. This solution is among the top 3 in the ICCV 2023 OmniObject3D Challenge.

cs.CV↗

arXiv2023

Joint 3D Deployment and Resource Allocation for UAV-assisted Integrated Communication and Localization

In this paper, we investigate an unmanned aerial vehicle (UAV)-assisted integrated communication and localization network in emergency scenarios where a single UAV is deployed as both an airborne base station (BS) and anchor node to assist ground BSs in communication and localization services. We formulate an optimization problem to maximize the sum communication rate of all users under localization accuracy constraints by jointly optimizing the 3D position of the UAV, and communication bandwidth and power allocation of the UAV and ground BSs. To address the intractable localization accuracy constraints, we introduce a new performance metric and geometrically characterize the UAV feasible deployment region in which the localization accuracy constraints are satisfied. Accordingly, we combine Gibbs sampling (GS) and block coordinate descent (BCD) techniques to tackle the non-convex joint optimization problem. Numerical results show that the proposed method attains almost identical rate performance as the meta-heuristic benchmark method while reducing the CPU time by 89.3%.

cs.NI↗

arXiv2022

Deployment Optimization of Dual-functional UAVs for Integrated Localization and Communication

In emergency scenarios, unmanned aerial vehicles (UAVs) can be deployed to assist localization and communication services for ground terminals. In this paper, we propose a new integrated air-ground networking paradigm that uses dual-functional UAVs to assist the ground networks for improving both communication and localization performance. We investigate the optimization problem of deploying the minimal number of UAVs to satisfy the communication and localization requirements of ground users. The problem has several technical difficulties including the cardinality minimization, the non-convexity of localization performance metric regarding UAV location, and the association between user and communication terminal. To tackle the difficulties, we adopt D-optimality as the localization performance metric, and derive the geometric characteristics of the feasible UAV hovering regions in 2D and 3D based on accurate approximation values. We solve the simplified 2D projection deployment problem by transforming the problem into a minimum hitting set problem, and propose a low-complexity algorithm to solve it. Through numerical simulations, we compare our proposed algorithm with benchmark methods. The number of UAVs required by the proposed algorithm is close to the optimal solution, while other benchmark methods require much more UAVs to accomplish the same task.

eess.SY↗

arXiv2021

Dynamic Trajectory and Offloading Control of UAV-enabled MEC under User Mobility

In this paper, we consider a UAV-enabled MEC platform that serves multiple mobile ground users with random movements and task arrivals. We aim to minimize the average weighted energy consumption of all users subject to the average UAV energy consumption and data queue stability constraints. To control the system operation in sequential time slots, we formulate the problem as a multi-stage stochastic optimization, and propose an online algorithm that optimizes the resource allocation and the UAV trajectory in each stage. We adopt Lyapunov optimization to convert the multi-stage stochastic problem into per-slot deterministic problems with much less optimizing variables. To tackle the non-convex per-slot problem, we use the successive convex approximation (SCA) technique to jointly optimize the resource allocation and the UAV movement. Simulation results show that the proposed online algorithm can satisfy the average UAV energy and queue stability constraints, and significantly outperform the other considered benchmark methods in reducing the energy consumption of ground users.

eess.SY↗