Institution profile

Harbin Engineering University

Academic institutionasia · cn
Official website
Research library163linked papers
Opportunities0open roles
Selected work

Representative Papers

ACW: Enhancing Traceability of AI-Generated Codes Based on Watermarking

Feb 12, 2024

Existing code watermarking methods for detecting large language model (LLM)-generated code suffer from poor generalizability, high computational overhead, and reliance on white-box access. To address these limitations, we propose ACW—a training-free, black-box, lightweight code watermarking framework. ACW embeds detectable watermarks implicitly during code generation via semantic-preserving and idempotent structured code transformations. It employs a statistically significant detection mechanism to robustly extract watermarks without requiring model internals or retraining. Evaluated across diverse LLMs—including ChatGPT and StarCoder—ACW achieves >98% detection accuracy on generated code and demonstrates strong resilience against common adversarial attacks such as code deletion, modification, and obfuscation. Notably, ACW is the first method to overcome the transferability bottleneck of text-based watermarks in the code domain. It simultaneously delivers zero training cost, broad model agnosticism, and high detection reliability.

2 citations1 influentialRead paper

The Benefits of Being Categorical Distributional: Uncertainty-aware Regularized Exploration in Reinforcement Learning

Oct 07, 2021

This paper investigates the theoretical advantages of distributional reinforcement learning (DRL) over classical RL, focusing on its implicit environmental exploration capability. Method: We provide the first rigorous decomposition of the distributional loss in categorical DRL, revealing an intrinsic, uncertainty-aware entropy regularization mechanism—spontaneously induced by the structure of the return distribution and requiring no explicit design. This adaptive regularizer transforms environmental uncertainty into enhanced reward signals for policy optimization. Unlike maximum-entropy RL, which explicitly encourages action-space diversity, this mechanism enables implicit, environment-driven exploration grounded in distributional shape. Contribution/Results: Our theoretical analysis uncovers the fundamental reason behind DRL’s superiority over classical RL. Empirical evaluation demonstrates that this implicit regularization significantly improves sample efficiency and policy robustness across diverse benchmarks.

2 citationsRead paper

SwarmFoam: An OpenFOAM Multi-Agent System Based on Multiple Types of Large Language Models

Jan 12, 2026

This work proposes SwarmFoam, the first large language model–based multi-agent framework that integrates multimodal perception, retrieval-augmented generation (RAG), and an intelligent error-correction mechanism to address the challenges of automating high-fidelity computational fluid dynamics (CFD) simulations in complex geometries. Traditional multi-agent systems struggle to effectively fuse multimodal inputs and achieve robust automation in such settings. SwarmFoam overcomes these limitations by jointly interpreting visual and natural language instructions to autonomously drive OpenFOAM simulations. Evaluated on 25 test cases, the system achieves an overall success rate of 84%, with 80% accuracy for natural language inputs and 86.7% for multimodal inputs, demonstrating significantly enhanced adaptability and automation capabilities for handling intricate geometries and diverse multimodal commands.

1 citationsRead paper
Recent publications

Latest Papers