Searcharxiv⌕ Search

arXiv subjects

Xingyu Wang

Publications and source records attributed to Xingyu Wang.

At least 19 recordsLinked to original sources
arXiv2026

HiLo-WAM: Decoupling Long-Horizon Task Evolution and Local Dynamics in World Action Models

World Action Models (WAMs) improve robotic manipulation by jointly modeling visual dynamics and robot actions. However, most existing WAMs model dense short-horizon visual dynamics, capturing local interactions but providing limited long-range task context. Extending prediction horizons increases uncertainty and computation while entangling task-level evolution with fine-grained local dynamics, obscuring their dependencies and distinct roles in action generation. We propose HiLo-WAM, which decouples long-horizon task evolution from local visual dynamics while jointly modeling their interaction with robot actions. A sparse long-horizon visual stream captures historical task context and future evolution, while a dense local dynamics visual stream models short-horizon robot-object interactions. The two streams operate at different temporal scales within a unified dual-DiT architecture and guide actions through structured cross-stream attention. Progressive inference further reduces unnecessary computation through stream-wise partial denoising, joint denoising, and dynamic skipping. Experiments demonstrate strong long-horizon manipulation performance: HiLo-WAM achieves a 93.70% average success rate on RoboTwin 2.0 and a 17.43 progress score on RoboDojo long-horizon tasks, with robust in-domain and OOD performance on the Fourier GR-3 robot. Ablations confirm the complementary contributions of long-horizon task evolution and local interaction dynamics, while temporal-scale analyses highlight the importance of balancing temporal coverage and resolution. Progressive inference further achieves a 2.93x speedup while improving performance.

cs.RO↗
arXiv2026

VL-Dedup: Graded Multimodal Entity Resolution for Image-Text Data Deduplication

Image-text corpora contain copies of one image, captions that describe one image in different ways, captions reused for different images, and captions that do not match their images. Each case calls for a different curation action. Deduplication and multimodal entity resolution (ER) only decide whether records are the same, and they are evaluated on labeled pairs. Yet pair scores do not tell which records end up in the same cluster or which useful captions get deleted. We therefore formulate image-text deduplication as graded multimodal ER, which decides visual identity, caption handling, and mismatches for each pair, and we evaluate it at three levels: pairs, clusters, and the retained set. Our method, VL-Dedup, groups records of the same visual entity into clusters and caps the size of each cluster. Within each cluster, a vision-language model (VLM) and CLIPScore decide which captions to keep and which to flag. We also build VL-DedupBench, with 50,000 COCO and SOP records and 66,788 labeled pairs. On it, a classifier that selects no false edge among the labeled pairs still joins millions of record pairs from different entities. Spending VLM calls on the pairs inside clusters, rather than on pairs with uncertain scores, gives better caption decisions from the same model. Even when almost all clusters are correct, keeping one record per cluster loses facts in many deletions, while the deletions of VL-Dedup lose much less information.

cs.CV↗
arXiv2026

BridgeWAM: Bridging Video and Action Experts via Latent Bridge Queries

World Action Models (WAMs) commonly couple video and action experts through dense, layer-wise Mixture-of-Transformers (MoT) attention, exposing action prediction to irrelevant visual features while enforcing matched expert depths. This coupling hinders selective visual-feature use and introduces redundant computation. We introduce Bridge-of-Experts (BoE), instantiated as BridgeWAM, which aggregates visual, language, and proprioceptive priors through 32 Latent Bridge Queries (LBQs) into a shared, limited latent space. Joint video-action supervision lets action gradients guide the Video Expert toward representations better suited for action prediction, while the Action Expert learns to selectively retrieve task-relevant information through cross-attention. Decoupling expert depths and reusing LBQs during action denoising yield a 5.06B model with a 30-layer Video Expert and a two-layer Action Expert, achieving 98.25% success on LIBERO and 72.24% pooled success on LIBERO-Plus, demonstrating effective task execution and strong generalization with an ultralight action head. Ablations support joint representation adaptation and unified LBQ conditioning. Relative to Fast-WAM, BridgeWAM reduces single-sample inference FLOPs by 28.72% in simulation profiling and real-robot latency to 77.1 ms (a 2.47-fold speedup). Qualitative experimental results further indicate preferential attention to target objects and task destinations across perturbations.

cs.RO↗
arXiv2026

Short-time existence and uniqueness for two-phase harmonic map heat flow with a prescribed moving interface

Let $N^+$ and $N^-$ be disjoint compact smooth submanifolds of a Euclidean space, and let $Λ\subset N^+\times N^-$ be a compact smooth embedded submanifold prescribing admissible pairs of one-sided traces. Given a smooth family of separating hypersurfaces $Γ_t\subset\mathbb T^d$, we study harmonic map heat flows into $N^\pm$ in the two moving phases, subject to \[ (u^+,u^-)\inΛ, \qquad \bigl(-\partial_{ν_t}u^+,\partial_{ν_t}u^-\bigr) \perp T_{(u^+,u^-)}Λ \quad\text{on }Γ_t. \] For $q>d+2$, initial data in $W_q^{2-2/q}$ satisfying these conditions at $t=0$ generate a unique short-time strong solution in $W_q^{2,1}$, which is smooth up to the moving interface for every positive time; smooth initial data satisfying the compatibility conditions of every order yield solutions that are smooth up to the initial corner. We double the paired map across $Λ$ by normal reflection, which turns the interface conditions into a Dirichlet problem near the interface, and match this problem with the two bulk phases by an overlapping space-time Schwarz map inside a Schauder fixed-point argument. When the interface evolves independently by mean curvature, we also obtain a blow-up alternative in terms of its curvature and the two phase gradients. As an application, we prove local smooth solvability for the periodic matrix-valued sharp-interface system appearing in the convergence theory of Fei, Lin, Wang, and Zhang (\emph{Invent. Math.} 233 (2023), 1--80).

math.AP↗
arXiv2026

JuZhou 1.0 Technical Report: The First Edge-Native Text-to-Image Foundation Model Trained Entirely on China-Developed AI Accelerators

Text-to-image (T2I) diffusion models typically require substantial computational resources and cloud infrastructure, posing significant challenges for edge deployment in terms of latency, cost, and user privacy. We present JuZhou 1.0, an ultra-lightweight T2I foundation model designed for fully offline, on-device execution. JuZhou 1.0 achieves its efficiency through four key designs: (1) a compact image-generation backbone consisting of a 0.385B-parameter denoising U-Net and a 1.90M-parameter distilled decoder, totaling approximately 0.387B parameters; (2) Rectified Flow training combined with DMD2 distillation, reducing inference to 4 sampling steps; (3) Chinese semantic alignment trained on 9M curated image-text pairs, enabling direct Chinese prompting without external translation at inference time; and (4) a training and distillation pipeline completed on domestically developed Sugon K100 AI accelerators without relying on NVIDIA GPUs for training or distillation. Despite its compact scale, the 28-step base model of JuZhou 1.0 achieves an overall GenEval score of 0.69, outperforming published baselines including SDXL (2.6B, 0.55), SD3-Medium (2B, 0.62), and IF-XL (4.3B, 0.61). We further validate the full poetry-to-image pipeline on Android and the core CLIP-U-Net-VAE generation branch on iOS. On a smartphone powered by the Snapdragon 8 Elite Gen 5 Mobile Platform, the 4-step U-Net denoising branch runs in approximately 1.6 seconds, while the full Android poetry-to-image pipeline takes 4.5 seconds with on-device prompt refinement on Xiaomi 17 Pro Max. These results position JuZhou 1.0 as a practical approach to mobile text-to-image generation and provide a concrete reference for Chinese-native generation, domestic-compute training, and fully offline on-device deployment after one-time installation.

cs.CV↗
arXiv2026

E-AVI: Evidence-Grounded Multimodal Assessment for Automated Video Interviews

Automated video interview assessment integrates verbal content, acoustic delivery, and visual behavior, yet numerical predictions alone provide limited inspectable support. We present E-AVI, an evidence-grounded framework that extracts timestamped multimodal evidence and integrates dimension-conditioned evidence attention with source-level embeddings for scoring. A shared evidence pool further supports natural-language feedback and follow-up question answering. On RecruitView and a private hospitality dataset, E-AVI consistently outperforms fine-tuned multimodal baselines in rank correlation. Ablation, evidence-deletion, bootstrap, human-audit, and QA analyses characterize the predictive contribution, grounding, and practical utility of the evidence pathway. Together, these results demonstrate that our proposed E-AVI framework improves predictive performance while providing inspectable support for assessment, feedback, and interactive analysis.

cs.AI↗
arXiv2026

Hull First, Wake Second: Wake-Reliance Suppression for Robust Maritime Vessel Detection

Maritime vessel detectors often face scenes where hulls are small, low-contrast, or blurred, while wakes are longer and easier to detect. This creates a wake-reliance problem: detectors may miss slow or stationary vessels with weak wakes, or produce false positives on wake-like water clutter. We propose HullWake, a hull-first wake-second framework for robust maritime vessel detection. HullWake separates proposal-centered hull evidence from directional wake context, extracts wake cues with bidirectional proposal-anchored corridors, and suppresses wake-dominant predictions through wake response supervision, wake-attenuated consistency, wake-only confidence suppression, and hull--wake decorrelation. We also introduce a wake-oriented evaluation protocol covering weak/no-wake vessels, wake-like hard negatives, worst-group AP, and confidence drop after wake attenuation. Experiments are conducted on Curated-Wake, a wake-oriented maritime dataset of about 10,000 images curated from Ships/Vessels in Aerial Images, the SMD benchmark, and SeaDronesSee, with newly added detection- and segmentation-level wake annotations. Compared with box-only detectors and mask-supervised segmentation baselines, HullWake improves overall AP, weak/no-wake robustness, wake-like false positives, worst-group AP, and confidence stability after wake attenuation.

cs.CV↗
arXiv2026

Primal-Dual Halpern-PAGE Algorithm for Constrained Stochastic Weakly Convex Optimization

We tackle the challenging problem of stochastic weakly convex optimization subject to mixed (equality and inequality) expected-value constraints. While optimal $\mathcal{O}(ε^{-3})$ sample complexity algorithms exist for unconstrained weakly convex problems, dealing with complex functional constraints typically requires cumbersome multi-loop penalty or augmented Lagrangian methods, which suffer from high inner-loop complexity and sensitive parameter tuning. To bridge this fundamental gap, we propose the primal-dual Halpern-PAGE (PD-HP) algorithm. As a purely single-loop method, PD-HP completely bypasses the computational burden of nested iterations. At each step, it merely requires solving a simple strongly convex surrogate subproblem alongside a straightforward dual projection, making it exceptionally efficient and convenient to implement. Crucially, we prove that this computationally lightweight algorithm achieves the optimal $\mathcal{O}(ε^{-3})$ sample complexity for mixed-constrained stochastic weakly convex problems, successfully matching the theoretical lower bounds. Furthermore, when the primal domain is a compact polyhedral convex set, we establish the deterministic stability of the dual multipliers by exploiting the generalized Mangasarian-Fromovitz constraint qualification (MFCQ) alongside Hoffman's error bound. This ensures that our optimal complexity bound holds strictly under the standard, unbounded KKT residual metric without any theoretical gaps or artificial residual truncations.

math.OC↗
arXiv2026

The sharp-interface limit of the matrix-valued Allen--Cahn equation

We study the sharp-interface limit of a matrix-valued Allen--Cahn equation with the Saint Venant--Kirchhoff potential \[ F(\mathbf{A})=\frac14\|{\mathbf{A}\mathbf{A}^\top-\mathbf{I}}\|^2 . \] The zero set of this potential is the orthogonal group $\mathbb{O}_n=\mathbb{O}_n^+\cup\mathbb{O}_n^-$, and the corresponding limiting problem combines mean-curvature motion of the interface with harmonic-map heat flow in the two bulk phases. The proof combines a modulated-energy argument with compactness estimates obtained from two skew-symmetric commutator formulations of the equation. The method avoids the spectral analysis of linearized operators around quasi-minimal connecting orbits and the construction of high-order matched asymptotic expansions. In particular, the limiting maps satisfy the minimal-pair condition on the moving interface and the weak transmission identities which, for smooth limits, are equivalent to the Neumann-type jump condition of the sharp-interface system.

math.AP↗
arXiv2026

UniScale: Adaptive Unified Inference Scaling via Online Joint Optimization of Model Routing and Test-Time Scaling

In real-world deployments of large language models (LLMs), balancing inference quality and computational cost has become a central challenge. Existing approaches tackle this trade-off along two largely independent dimensions: model routing, which switches among models of different scales to match request complexity, and test-time scaling (TTS), which adjusts inference-time compute within a fixed model for fine-grained control. However, this decoupled design introduces inherent limitations. Model routing yields coarse-grained, discrete performance changes due to the sparse set of model scales, while single-model TTS often encounters capacity ceilings and exhibits diminishing returns as compute increases. Moreover, treating the two mechanisms separately restricts adaptability in dynamic inference environments. To overcome these limitations, we introduce Unified Inference Scaling (UIS), which unifies model routing and TTS in a single optimization space. Building on this formulation, we propose UniScale, an online framework that models adaptive UIS as a contextual multi-armed bandit problem and learns inference policies via LinUCB. The framework incorporates efficiency-aware learning and cost modeling to ensure stable and scalable optimization over high-dimensional action spaces. Evaluation shows that UniScale effectively exploits the synergy in the UIS space to deliver a fine-grained and consistently better quality-cost trade-off across diverse, dynamic inference scenarios.

cs.AI↗
arXiv2026

CUBic: Coordinated Unified Bimanual Perception and Control Framework

Recent advances in visuomotor policy learning have enabled robots to perform control directly from visual inputs. Yet, extending such end-to-end learning from single-arm to bimanual manipulation remains challenging due to the need for both independent perception and coordinated interaction between arms. Existing methods typically favor one side -- either decoupling the two arms to avoid interference or enforcing strong cross-arm coupling for coordination -- thus lacking a unified treatment. We propose CUBic, a Coordinated and Unified framework for Bimanual perception and control that reformulates bimanual coordination as a unified perceptual modeling problem. CUBic learns a shared tokenized representation bridging perception and control, where independence and coordination emerge intrinsically from structure rather than from hand-crafted coupling. Our approach integrates three components: unidirectional perception aggregation, bidirectional perception coordination through two codebooks with shared mapping, and a unified perception-to-control diffusion policy. Extensive experiments on the RoboTwin benchmark show that CUBic consistently surpasses standard baselines, achieving marked improvements in coordination accuracy and task success rates over state-of-the-art visuomotor baselines.

cs.RO↗
arXiv2026

Prox-PEP: A Proximal Partial Exact Penalty Algorithm for Weakly Convex Stochastic Nonlinear Programming

This paper considers stochastic optimization problems with weakly convex objective and constraint functions. We propose Prox-PEP, a proximal method equipped with quadratic subproblems. To handle nonlinear equality constraints, we employ an exact penalty approach, transforming them into inequality constraints with auxiliary slack variables. At each iteration, we construct quadratic approximations for both the objective and the constraint functions to facilitate efficient subproblem computation. By carefully designing the second-order approximation matrices, the subproblem constructed via the augmented Lagrangian function is strictly guaranteed to be strongly convex. Furthermore, we adopt a dynamic strategy for the equality penalty parameter: it monotonically increases up to a predefined threshold and remains constant thereafter. Building upon this algorithmic framework, we establish comprehensive asymptotic complexities. We prove that Prox-PEP achieves an $\mathcal{O}(T^{-1/4})$ average expected oracle complexity for $ε$-KKT stationarity, specifically bounding the squared norm of the gradient of the Moreau envelope of the Lagrangian function, alongside constraint violations and complementarity conditions. Additionally, under standard light-tailed martingale noise assumptions, we derive an $\mathcal{O}(T^{-1/8})$ high-probability convergence bound for the norm of the gradient of the Lagrangian's Moreau envelope, as well as $\mathcal{O}(T^{-1/4})$ high-probability bounds for both constraint violations and complementarity conditions.

math.OC↗
arXiv2026

Semantics Over Syntax: Uncovering Pre-Authentication 5G Baseband Vulnerabilities

Modern 5G user equipment (UE) processes Radio Resource Control (RRC) configuration messages during early control-plane exchanges, before authentication and integrity protection are established. Prior work for testing 5G UEs has largely focused on constructing syntactically invalid inputs. In contrast, we show that syntactically valid but semantically inconsistent messages, which violate specification-level field constraints or cross-field dependencies, can drive baseband implementations into invalid states, triggering assertion failures or modem crashes. These findings reveal semantic inconsistencies in pre-authentication signaling as a critical yet underexplored attack surface in 5G UE implementations. To address this gap, we present Constraint-Guided Semantic Testing (ConSeT), a framework that systematically extracts specification-level constraints and leverages them to generate targeted semantic violations for testing 5G UEs. ConSeT decodes RRC messages into structured fields, derives schema-based rules, infers cross-field dependencies using a Large Language Model (LLM) in an evidence-bounded manner, and produces syntactically valid test cases that intentionally violate semantic constraints. We evaluate ConSeT on both commercial and open-source 5G UEs. On commercial smartphones, it uncovers 7 previously unknown vulnerabilities through responsible disclosure, including 3 high-severity CVEs, affecting 64 chipset models and over 542 commercially available smartphone models. On the open-source OAI UE, ConSeT additionally triggers 29 distinct crash sites.

cs.CR↗
arXiv2026

FOZO: Forward-Only Zeroth-Order Prompt Optimization for Test-Time Adaptation

Test-Time Adaptation (TTA) is essential for enabling deep learning models to handle real-world data distribution shifts. However, current approaches face significant limitations: backpropagation-based methods are not suitable for low-end deployment devices, due to their high computation and memory requirements, as well as their tendency to modify model weights during adaptation; while traditional backpropagation-free techniques exhibit constrained adaptation capabilities. In this work, we propose Forward-Only Zeroth-Order Optimization (FOZO), a novel and practical backpropagation-free paradigm for TTA. FOZO leverages a memory-efficient zeroth-order prompt optimization, which is led by objectives optimizing both intermediate feature statistics and prediction entropy. To ensure efficient and stable adaptation over the out-of-distribution data stream, we introduce a dynamically decaying perturbation scale during zeroth-order gradient estimation and theoretically prove its convergence under the TTA data stream assumption. Extensive continual adaptation experiments on ImageNet-C, ImageNet-R, and ImageNet-Sketch demonstrate FOZO's superior performance, achieving 59.52% Top-1 accuracy on ImageNet-C (5K, level 5) and outperforming main gradient-based methods and SOTA forward-only FOA (58.13%). Furthermore, FOZO exhibits strong generalization on quantized (INT8) models. These findings demonstrate that FOZO is a highly competitive solution for TTA deployment in resource-limited scenarios.

cs.CV↗
arXiv2026

Uncovering bistability phenomena in two-layer Couette flow experiments using nonlocal evolution equations

This paper investigates the stability of interfacial long waves in two-layer plane Couette flow using a nonlinear, nonlocal asymptotic model derived from the Navier-Stokes equations and valid for thin upper layers. Nonlocality enters through a coupling of the thin and main layers, and crucial inertial effects are retained. The models generically support bistability phenomena observed in experiments where two stable travelling waves, one unimodal and the other bimodal, are recorded at the same lid velocity. In direct comparisons with experiments, the models show remarkable agreement, both qualitatively and quantitatively. The two stable travelling waves are identified and their basins of attraction characterised via large-time computations for different initial conditions. We also identify a new symmetry-breaking travelling-wave branch bifurcating from the bimodal family, compute higher-wavenumber travelling-wave branches, and present time-periodic orbits arising via Hopf bifurcation.

physics.flu-dyn↗
arXiv2026

First-Exit Time Analysis for Truncated Heavy-Tailed Dynamical Systems

In this paper, we study the first-exit time of stochastic difference equation $X^η_{j+1}(x) = X^η_{j}(x) + ηa\big( X^η_{j}(x)\big) + ησ\big( X^η_{j}(x)\big)Z_{j+1}$ and its truncated variant $X^{η|b}_{j+1}(x) = X^{η|b}_{j}( x) + φ_b\big(ηa\big( X^{η|b}_{j}( x)\big) + ησ\big( X^{η|b}_{j}( x)\big) Z_{j+1}\big)$, where $φ_b(x) = (x/|x|)\min\{|x|, b\}$ and the law of the noise $Z_t$ is multivariate regularly varying. The truncation operator $φ_b(\cdot)$ is often introduced as a modulation mechanism in heavy-tailed systems, such as stochastic gradient descent algorithms in deep learning. By developing a framework that connects large deviations with metastability, we leverage the locally uniform sample-path large deviations for both processes in Wang and Rhee (2024) to obtain precise characterizations of the joint distributions of the first exit times and exit locations. The resulting limit theorem unveils a discrete hierarchy of phase transitions (i.e., exit times) as the truncation threshold $b$ varies, and manifests the catastrophe principle, whereby key events or metastable behaviors in heavy-tailed systems are driven by catastrophic behavior in a few components while the rest of the system behaves nominally. These developments lead to a comprehensive heavy-tailed counterpart of the classical Freidlin-Wentzell theory.

math.PR↗
arXiv2025

GroupDebate: Enhancing the Efficiency of Multi-Agent Debate Using Group Discussion

In recent years, Large Language Models (LLMs) have demonstrated remarkable capabilities across diverse NLP tasks. Extensive research has explored how to enhance the logical reasoning abilities such as Chain-of-Thought, Chain-of-Thought with Self-Consistency, Tree-Of-Thoughts, and multi-agent debates. In the context of multi-agent debates, significant performance improvements can be achieved with an increasing number of agents and debate rounds. However, the escalation in the number of agents and debate rounds can drastically raise the tokens cost of debates, thereby limiting the scalability of the multi-agent debate technique. To better harness the advantages of multi-agent debates in logical reasoning tasks, this paper proposes a method to significantly reduce token cost in multi-agent debates. This approach involves dividing all agents into multiple debate groups, with agents engaging in debates within their respective groups and sharing interim debate results between groups. Comparative experiments across multiple datasets have demonstrated that this method can reduce the total tokens by up to 51.7% during debates and while potentially enhancing accuracy by as much as 25%. Our method significantly enhances the performance and efficiency of interactions in the multi-agent debate.

cs.CL↗
arXiv2025

TCC-Bench: Benchmarking the Traditional Chinese Culture Understanding Capabilities of MLLMs

Recent progress in Multimodal Large Language Models (MLLMs) have significantly enhanced the ability of artificial intelligence systems to understand and generate multimodal content. However, these models often exhibit limited effectiveness when applied to non-Western cultural contexts, which raises concerns about their wider applicability. To address this limitation, we propose the Traditional Chinese Culture understanding Benchmark (TCC-Bench), a bilingual (i.e., Chinese and English) Visual Question Answering (VQA) benchmark specifically designed for assessing the understanding of traditional Chinese culture by MLLMs. TCC-Bench comprises culturally rich and visually diverse data, incorporating images from museum artifacts, everyday life scenes, comics, and other culturally significant contexts. We adopt a semi-automated pipeline that utilizes GPT-4o in text-only mode to generate candidate questions, followed by human curation to ensure data quality and avoid potential data leakage. The benchmark also avoids language bias by preventing direct disclosure of cultural concepts within question texts. Experimental evaluations across a wide range of MLLMs demonstrate that current models still face significant challenges when reasoning about culturally grounded visual content. The results highlight the need for further research in developing culturally inclusive and context-aware multimodal systems. The code and data can be found at: https://tcc-bench.github.io/.

cs.MM↗