Searcharxiv⌕ Search

arXiv subjects

Hua Yang

Publications and source records attributed to Hua Yang.

At least 19 recordsLinked to original sources
arXiv2026

TTPO: Test-Time Policy Optimization

Recent prominent post-training methods, such as Reinforcement Learning (RL) and On-Policy Self-Distillation (OPSD), have driven rapid progress in mathematical reasoning for large language models, yet their reliance on ground-truth labels precludes test-time training (TTT). Replacing ground truth with majority-vote pseudo-labels is a natural alternative, yet it is fragile: an incorrect vote corrupts the teacher and misleads every token. We observe that this failure mode is asymmetric: rollouts that disagree with the pseudo-label are typically wrong regardless of whether the vote itself is correct. Building on this observation, we propose Test-Time Policy Optimization (TTPO), an asymmetric objective that distills agreeing rollouts via OPSD and penalizes disagreeing rollouts with Grouped RL. Token-level selection further refines both branches: distillation down-weights already-converged positions, while RL penalizes only confident errors. Both updates remain well-grounded even under frequent pseudo-label errors, and majority-vote routing yields tighter self-supervision as the model improves. Without any labels, TTPO matches label-supervised OPSD on five competition-level benchmarks, raises Qwen3-1.7B from 38.0% to 45.2% in TTT, yields +25.2% to +36.4% without thinking, and shows strong cross-task generalization.

cs.CL↗
arXiv2026

DaViNCi: A Dataset Towards Outdoor Vision-and-Language Navigation with Continuous Actions and Dynamic Elements

Vision-and-Language Navigation (VLN) has progressively expanded from indoor to outdoor environments. However, existing outdoor VLN datasets still rely on fixed discrete topological graphs for construction. It fails to align with the rapidly changing real-world outdoor environments and impedes the sim-to-real transfer of VLN agents. To address this limitation, we propose DaViNCi (\textbf{D}yn\textbf{a}mic \textbf{Vi}sion-and-Language \textbf{N}avigation in \textbf{C}ont\textbf{i}nuous Environment), the first outdoor VLN dataset that simultaneously introduces both continuous and dynamic factors. The agent not only moves in the outdoor environment using continuous actions but is also required to handle unpredictable dynamic elements. The dataset encompasses six distinct maps with a total of 6,933 trajectories. Through comprehensive comparative experiments, we find that the success rate on DaViNCi decreased by more than 10\% in discrete environments compared to previous datasets. And there is an even greater decline in continuous settings, demonstrating the challenge of DaViNCi. Furthermore, we clarify the impact of action granularity and dynamic elements. These results demonstrate the practical value of DaViNCi in advancing outdoor VLN toward more realistic environments. The website is https://xzh0312.github.io/DaViNCi/.

cs.RO↗
arXiv2026

FasTac: A Curved Multispectral Vision-Based Tactile Sensor for High-Speed High-Precision 3D Shape and Force Perception

Curved tactile fingertips for dexterous manipulation must resolve fine contact geometry, distinguish normal and tangential loads, and capture transient signals. Existing curved vision-based tactile sensors struggle to combine accurate 3D reconstruction, three-axis force estimation, and high-speed processing in a compact form. This article presents FasTac, a curved vision-based tactile sensor integrating multispectral photometric stereo, dynamic-convolution force estimation, and hardware acceleration on a field-programmable gate array (FPGA). Single-image-sensor simultaneous multispectral imaging provides spatially aligned observations for robust surface normal estimation, followed by boundary-prior fast Poisson depth reconstruction. HyperForce uses position-aware dynamic convolution to model the spatially nonuniform mechanical response of curved elastomers and estimate three-axis forces. The complete image-to-normal-force pipeline is deployed on an FPGA. Experiments show that near-infrared (NIR) illumination and the boundary prior decrease depth mean absolute error (MAE) from 0.2730 mm to 0.0415 mm; HyperForce achieves normalized mean absolute error (NMAE) values of 2.74% and 2.39% for normal and shear forces, respectively; and FPGA deployment shortens processing latency from 3.26 ms on the GPU to 1.09 ms. Multi-object reconstruction, feedback grasping, and vibration measurement validate fine geometric perception, stable force feedback, and dynamic contact sensing.

cs.RO↗
arXiv2026

SERPO: Self-Evolving Rubric Policy Optimization for Open-Ended Test-Time Reinforcement Learning

Test-time reinforcement learning (TTRL) enables language models to self-evolve at inference time without labeled feedback. Existing methods rely on answer voting and therefore do not extend naturally to open-ended generation, where valid responses cannot be mapped to a shared canonical answer. Without external reward models or stronger judges, adaptation must instead construct reliable rewards from the model's own outputs. We introduce SERPO (Self-Evolving Rubric Policy Optimization), which replaces answer voting with a closed loop that co-evolves response evidence, query-specific rubrics, and policy parameters. Good-Normal-Bad (G-N-B) response evolution organizes maximally separated rollouts into ordered archives; rubric evolution retains criteria that discriminate these archives; probabilistic criterion scoring converts verdict-token likelihoods into reward signals; and policy evolution optimizes the actor with the resulting signals. New actor rollouts then refresh both the archives and rubrics, closing the three-way evolution loop. Across two model configurations, two in-domain benchmarks, and four OOD benchmarks, SERPO improves HealthBench and ResearchQA by up to 20.63 and 20.31 points over the corresponding base models, raises the six-benchmark macro-average by up to 8.06 points, and supports OOD transfer and continued cross-benchmark evolution.

cs.CL↗
arXiv2026

Chronos: A Physics-Informed Full-History Framework for Non-Markovian Long-Horizon Manipulation

General-purpose robot policies should be modeled as dynamical systems, yet many VLA and generative imitation policies still rely on present observations or short windows. This Markovian shortcut fails in memory-dependent manipulation: identical observations can demand different actions after different histories. We present Chronos, a physics-informed full-history framework for non-Markovian long-horizon manipulation. The key idea is to elevate observation history from auxiliary context to the latent state of the policy dynamics. At each physical control step, Chronos forms one state-representative token by fusing observation and proprioception, so the token sequence is aligned one-to-one with physical time. A selective state space model propagates this causal historical state, which conditions a multimodal coarse action prior through implicit maximum likelihood estimation (IMLE). This prior is then refined by a second-order Schrodinger-inspired bridge that predicts acceleration fields, yielding smoother and more physically grounded robot motion. Across 16 simulated tasks and 4 real-world experiments, Chronos is evaluated on precision insertion, general manipulation, and memory-dependent long-horizon control. On RMBench, where success requires remembering task phase, Chronos achieves 73.6% average success, outperforming Markovian VLA baseline pi0.5 by +62.4 percentage points, a 6.6x relative gain, while using 10x fewer parameters. It also surpasses the memory VLA Mem-0 by 22.8 points while using over 30x fewer parameters. In real-world dual-arm experiments using a single RGB camera, Chronos achieves 78% average success over four tasks, including 72% on the three memory-dependent tasks, whereas pi0.5 achieves 7% overall and 0% on the memory-dependent subset. These results suggest that history should not be treated as auxiliary context, but as the latent state of the manipulation policy.

cs.RO↗
arXiv2026

SpaceVLN: A Zero-Shot Vision-and-Language Navigation Agent with Online Spatial Cognitive Memory and Reasoning

Vision-and-Language Navigation in continuous environments requires agents to understand the spatial structure of previously unseen environments in order to follow language instructions. Although foundation models have opened a promising path toward zero-shot navigation without task-specific policy training, many navigators still rely on local visual cues and linear history-based reasoning, overlooking the spatial nature of navigation across explored regions, traversed paths, landmarks, and their spatial relations. In this paper, we propose SpaceVLN, a navigation agent built around Spatial Cognitive Memory and Task-Guided Spatial Reasoning. Specifically, SpaceVLN introduces an efficient stagewise closed-loop framework where planning and execution are organized around verifiable space--landmark stages. During navigation, the agent progressively abstracts explored regions into Spatial Waypoints and dynamically maintains subtask-grounded landmark evidence, forming a hierarchical Spatial Cognitive Memory for progress localization and spatial-relation understanding. Built on this memory, Spatial-CoT integrates task-progress reasoning with spatial perception, analysis, and prediction, enabling Task-Guided Spatial Reasoning for embodied navigation. The unified stage interface enables SpaceVLN to address both Vision-and-Language Navigation and Object-Goal Navigation under a unified zero-shot setting, without task-specific policy training. Across R2R-CE, RxR-CE, GN-Bench, and HM3D-OVON, SpaceVLN achieves state-of-the-art zero-shot performance, and real-robot deployment further validates its applicability. These results highlight Spatial Cognitive Memory and Task-Guided Spatial Reasoning as a practical foundation for stronger embodied navigation agents.

cs.RO↗
arXiv2026

MAD-OPD: Breaking the Ceiling in On-Policy Distillation via Multi-Agent Debate

On-policy distillation (OPD) trains a student on its own trajectories under token-level teacher supervision, but existing methods are capped by a single-teacher capability ceiling: when the teacher errs, the student inherits the error. OPD also remains largely unexplored in agentic tasks, where per-step errors compound across long trajectories and destabilize training. We propose MAD-OPD (Multi-Agent Debate-driven On-Policy Distillation), which breaks this ceiling by recasting the distillation teacher as a deliberative collective of teachers that debate over the student's on-policy state; the debate produces an emergent collective intelligence that supplies token-level supervision, with each teacher's contribution weighted by its post-debate confidence. To extend OPD to agentic tasks, we also introduce On-Policy Agentic Distillation (OPAD), which adds step-level sampling to stabilize training under multi-step error compounding. We additionally derive a task-adaptive divergence principle, selecting JSD (Jensen-Shannon divergence) for agentic stability and reverse KL (Kullback-Leibler) divergence for code generation, and verify it both theoretically and empirically. Across six teacher-student configurations (Qwen3 and Qwen3.5; 1.7B-14B students, 8B-32B teachers) and five agentic and code benchmarks, MAD-OPD ranks first across all six configurations; on the 14B+8B$\to$4B setting it lifts the agentic average by $+2.4\%$ and the code average by $+3.7\%$ over the stronger single-teacher OPD.

cs.CL↗
arXiv2026

An in situ self-adaptive hydrogel coating enables seamless neural interfaces via okra mucilage polysaccharide and {\alpha}-helical peptide amphiphiles co-assembly

Long-term stability of neural interfaces is frequently compromised by mechanical mismatch and chronic neuroinflammation, often leading to electrode detachment and signal failure. While hydrogel coatings offer a solution, conventional designs typically rely on exogenous conductive fillers that can sacrifice mechanical flexibility or induce toxicity. Here, we report on a soft neural interface based on the supramolecular co-assembly of a renewable natural polysaccharide, okra mucilage polysaccharide (OMP), and an {\alpha}-helical peptide amphiphiles (APA). The resulting OMP-APA hydrogel (OP gel) exhibits environment-responsive enhancements in bioadhesion and charge-transport capability triggered by physiological pH and electrical stimulation. These properties arise from intrinsic, stimulus-responsive alterations in fibre architecture and orientation, eliminating the need for conductive fillers. Leveraging interfacial liquid-liquid phase separation, we demonstrate the in situ coating of ultra-thin OP-gel coating onto carbon fibre electrodes (CFE). The OP-gel-coated electrodes (OP-CFE) significantly mitigate foreign body responses and glial scarring, enabling stable, high-quality neural recordings in a mouse cortical in vivo model. Our findings provide a versatile strategy for constructing seamless, multifunctional bio-interfaces through supramolecular co-assembly, with broad implications for advancing neural prosthetics and neuroscience research.

physics.bio-ph↗
arXiv2026

Adaptive low-rank exponential integrators for large-scale differential Riccati equation

Matrix differential Riccati equation (DRE) typically exhibits transient and steady-state phases, posing challenges for fixed-step time integration methods, which may lack accuracy during transients or oversample in steady regimes. In this work, we propose adaptive low-rank matrix-valued exponential integrators for large-scale stiff DRE. The methods combine embedded exponential Rosenbrock-type schemes and adaptive step-size control, enabling an automatic adjustment to the evolving solution dynamics. This improves the accuracy during rapid transient phases while maintaining high accuracy in the steady state. Numerical experiments on benchmark problems demonstrate that the proposed adaptive integrators consistently improve accuracy and computational efficiency compared with fixed-step low-rank schemes.

math.NA↗
arXiv2026

P-GenRM: Personalized Generative Reward Model with Test-time User-based Scaling

Personalized alignment of large language models seeks to adapt responses to individual user preferences, typically via reinforcement learning. A key challenge is obtaining accurate, user-specific reward signals in open-ended scenarios. Existing personalized reward models face two persistent limitations: (1) oversimplifying diverse, scenario-specific preferences into a small, fixed set of evaluation principles, and (2) struggling with generalization to new users with limited feedback. To this end, we propose P-GenRM, the first Personalized Generative Reward Model with test-time user-based scaling. P-GenRM transforms preference signals into structured evaluation chains that derive adaptive personas and scoring rubrics across various scenarios. It further clusters users into User Prototypes and introduces a dual-granularity scaling mechanism: at the individual level, it adaptively scales and aggregates each user's scoring scheme; at the prototype level, it incorporates preferences from similar users. This design mitigates noise in inferred preferences and enhances generalization to unseen users through prototype-based transfer. Empirical results show that P-GenRM achieves state-of-the-art results on widely-used personalized reward model benchmarks, with an average improvement of 2.31%, and demonstrates strong generalization on an out-of-distribution dataset. Notably, Test-time User-based scaling provides an additional 3% boost, demonstrating stronger personalized alignment with test-time scalability.

cs.CL↗
arXiv2026

SAL: Selective Adaptive Learning for Backpropagation-Free Training with Sparsification

Standard deep learning relies on Backpropagation (BP), which is constrained by biologically implausible weight symmetry and suffers from significant gradient interference within dense representations. To mitigate these bottlenecks, we propose Selective Adaptive Learning (SAL), a training method that combines selective parameter activation with adaptive area partitioning. Specifically, SAL decomposes the parameter space into mutually exclusive, sample-dependent regions. This decoupling mitigates gradient interference across divergent semantic patterns and addresses explicit weight symmetry requirements through our refined feedback alignment. Empirically, SAL demonstrates competitive convergence rates, leading to improved classification performance across 10 standard benchmarks. Additionally, SAL achieves numerical consistency and competitive accuracy even in deep regimes (up to 128 layers) and large-scale models (up to 1B parameters). Our approach is loosely inspired by biological learning mechanisms, offering a plausible alternative that contributes to the study of scalable neural network training.

cs.LG↗
arXiv2025

How Do Semantically Equivalent Code Transformations Impact Membership Inference on LLMs for Code?

The success of large language models for code relies on vast amounts of code data, including public open-source repositories, such as GitHub, and private, confidential code from companies. This raises concerns about intellectual property compliance and the potential unauthorized use of license-restricted code. While membership inference (MI) techniques have been proposed to detect such unauthorized usage, their effectiveness can be undermined by semantically equivalent code transformation techniques, which modify code syntax while preserving semantic. In this work, we systematically investigate whether semantically equivalent code transformation rules might be leveraged to evade MI detection. The results reveal that model accuracy drops by only 1.5% in the worst case for each rule, demonstrating that transformed datasets can effectively serve as substitutes for fine-tuning. Additionally, we find that one of the rules (RenameVariable) reduces MI success by 10.19%, highlighting its potential to obscure the presence of restricted code. To validate these findings, we conduct a causal analysis confirming that variable renaming has the strongest causal effect in disrupting MI detection. Notably, we find that combining multiple transformations does not further reduce MI effectiveness. Our results expose a critical loophole in license compliance enforcement for training large language models for code, showing that MI detection can be substantially weakened by transformation-based obfuscation techniques.

cs.SE↗
arXiv2025

Understanding Privacy Risks in Code Models Through Training Dynamics: A Causal Approach

Large language models for code (LLM4Code) have greatly improved developer productivity but also raise privacy concerns due to their reliance on open-source repositories containing abundant personally identifiable information (PII). Prior work shows that commercial models can reproduce sensitive PII, yet existing studies largely treat PII as a single category and overlook the heterogeneous risks among different types. We investigate whether distinct PII types vary in their likelihood of being learned and leaked by LLM4Code, and whether this relationship is causal. Our methodology includes building a dataset with diverse PII types, fine-tuning representative models of different scales, computing training dynamics on real PII data, and formulating a structural causal model to estimate the causal effect of learnability on leakage. Results show that leakage risks differ substantially across PII types and correlate with their training dynamics: easy-to-learn instances such as IP addresses exhibit higher leakage, while harder types such as keys and passwords leak less frequently. Ambiguous types show mixed behaviors. This work provides the first causal evidence that leakage risks are type-dependent and offers guidance for developing type-aware and learnability-aware defenses for LLM4Code.

cs.SE↗
arXiv2025

DLADiff: A Dual-Layer Defense Framework against Fine-Tuning and Zero-Shot Customization of Diffusion Models

With the rapid advancement of diffusion models, a variety of fine-tuning methods have been developed, enabling high-fidelity image generation with high similarity to the target content using only 3 to 5 training images. More recently, zero-shot generation methods have emerged, capable of producing highly realistic outputs from a single reference image without altering model weights. However, technological advancements have also introduced significant risks to facial privacy. Malicious actors can exploit diffusion model customization with just a few or even one image of a person to create synthetic identities nearly identical to the original identity. Although research has begun to focus on defending against diffusion model customization, most existing defense methods target fine-tuning approaches and neglect zero-shot generation defenses. To address this issue, this paper proposes Dual-Layer Anti-Diffusion (DLADiff) to defense both fine-tuning methods and zero-shot methods. DLADiff contains a dual-layer protective mechanism. The first layer provides effective protection against unauthorized fine-tuning by leveraging the proposed Dual-Surrogate Models (DSUR) mechanism and Alternating Dynamic Fine-Tuning (ADFT), which integrates adversarial training with the prior knowledge derived from pre-fine-tuned models. The second layer, though simple in design, demonstrates strong effectiveness in preventing image generation through zero-shot methods. Extensive experimental results demonstrate that our method significantly outperforms existing approaches in defending against fine-tuning of diffusion models and achieves unprecedented performance in protecting against zero-shot generation.

eess.IV↗
arXiv2025

Adapter Shield: A Unified Framework with Built-in Authentication for Preventing Unauthorized Zero-Shot Image-to-Image Generation

With the rapid progress in diffusion models, image synthesis has advanced to the stage of zero-shot image-to-image generation, where high-fidelity replication of facial identities or artistic styles can be achieved using just one portrait or artwork, without modifying any model weights. Although these techniques significantly enhance creative possibilities, they also pose substantial risks related to intellectual property violations, including unauthorized identity cloning and stylistic imitation. To counter such threats, this work presents Adapter Shield, the first universal and authentication-integrated solution aimed at defending personal images from misuse in zero-shot generation scenarios. We first investigate how current zero-shot methods employ image encoders to extract embeddings from input images, which are subsequently fed into the UNet of diffusion models through cross-attention layers. Inspired by this mechanism, we construct a reversible encryption system that maps original embeddings into distinct encrypted representations according to different secret keys. The authorized users can restore the authentic embeddings via a decryption module and the correct key, enabling normal usage for authorized generation tasks. For protection purposes, we design a multi-target adversarial perturbation method that actively shifts the original embeddings toward designated encrypted patterns. Consequently, protected images are embedded with a defensive layer that ensures unauthorized users can only produce distorted or encrypted outputs. Extensive evaluations demonstrate that our method surpasses existing state-of-the-art defenses in blocking unauthorized zero-shot image synthesis, while supporting flexible and secure access control for verified users.

cs.CV↗
arXiv2025

A dispersal recolonisation 3D biofilm in vitro model based on co-assembled peptide amphiphiles and clinical wound fluid

Chronic wound infections are sustained by dynamic 3D biofilm cycles involving maturation, dispersal, and recolonisation, yet existing in vitro models fail to reproduce these temporal and structural complexities. Here, we report a strategy that co-assembles a designed protease-inhibitory peptide amphiphile (PA-GF) with patient-derived wound fluid (WF) to reconstruct the complete biofilm life cycle in vitro. The PA-GF sequence incorporates an HWGF motif capable of binding and inhibiting matrix metalloproteinase-9 (MMP-9), thereby preserving the integrity of recolonised biofilms under proteolytic stress. Co-assembling with WF generated a living material that faithfully mimicked the biochemical and mechanical microenvironment of chronic wounds, supporting the formation of stable 3D biofilms capable of dispersal and recolonisation. Furthermore, we established a controllable polymicrobial infection model and validated its translational relevance through antibiotic susceptibility profiling and spatial microbiological analyses. Notably, the antibiotic response patterns of the PA/WF-derived biofilms closely mirrored those observed in a rat wound infection in vivo model. Collectively, our findings demonstrate that co-assembling living materials can recapitulate the nutritional composition, 3D architecture, and recolonisation dynamics of in vivo infectious biofilms, offering a physiologically relevant and customisable platform for investigating chronic wound infections and accelerating anti-biofilm drug discovery.

physics.med-ph↗
arXiv2025

MM-HELIX: Boosting Multimodal Long-Chain Reflective Reasoning with Holistic Platform and Adaptive Hybrid Policy Optimization

While current Multimodal Large Language Models (MLLMs) have demonstrated proficiency in reasoning tasks such as mathematics and logic, their capacity for long-chain reflective reasoning, a prerequisite for solving complex real-world problems, remains largely underexplored. In this work, we first conduct an extensive empirical investigation to evaluate this capability. Leveraging a carefully designed data synthesis engine, we construct MM-HELIX, a multimodal benchmark consisting 1,260 samples of 42 challenging synthetic tasks that require iterative thinking and backtracking. Empirical results on this benchmark reveal that existing MLLMs exhibit significant performance deficits in long-chain reflective reasoning. To address this limitation, we generate post-training data and further explore learning paradigms for exploiting such data. We first develop the Step-Elicited Response Generation pipeline to create MM-HELIX-100K, a large-scale dataset of 100k high-quality, reflective reasoning traces for instruction-tuning stage. Given that standard Reinforcement Learning fails on complex tasks due to sparse reward signals and catastrophic forgetting after Supervised Fine-Tuning, we propose Adaptive Hybrid Policy Optimization (AHPO), a novel training strategy that dynamically unifies offline supervision and online optimization into a single stage. This strategy enables the model to learn from expert data when rewards are sparse and conduct independent exploration once proficient. When applied to the Qwen2.5-VL-7B baseline, our method achieves a +18.6\% accuracy improvement on MM-HELIX benchmark and demonstrates strong generalization with a +5.7\% average performance gain on general mathematic and logic tasks. Our work demonstrate that reflective reasoning in MLLMs can be effectively learned and generalized, paving the way for developing more capable MLLMs.

cs.CV↗
arXiv2025

How Quantization Impacts Privacy Risk on LLMs for Code?

Large language models for code (LLMs4Code) rely heavily on massive training data, including sensitive data, such as cloud service credentials of the projects and personal identifiable information of the developers, raising serious privacy concerns. Membership inference (MI) has recently emerged as an effective tool for assessing privacy risk by identifying whether specific data belong to a model's training set. In parallel, model compression techniques, especially quantization, have gained traction for reducing computational costs and enabling the deployment of large models. However, while quantized models still retain knowledge learned from the original training data, it remains unclear whether quantization affects their ability to retain and expose privacy information. Answering this question is of great importance to understanding privacy risks in real-world deployments. In this work, we conduct the first empirical study on how quantization influences task performance and privacy risk simultaneously in LLMs4Code. To do this, we implement widely used quantization techniques (static and dynamic) to three representative model families, namely Pythia, CodeGen, and GPTNeo. Our results demonstrate that quantization has a significant impact on reducing the privacy risk relative to the original model. We also uncover a positive correlation between task performance and privacy risk, indicating an underlying tradeoff. Moreover, we reveal the possibility that quantizing larger models could yield better balance than using full-precision small models. Finally, we demonstrate that these findings generalize across different architectures, model sizes and MI methods, offering practical guidance for safeguarding privacy when deploying compressed LLMs4Code.

cs.SE↗