Chen Lyu (吕晨)

Associate Professor, School of Information Science & Engineering
Shandong Normal University

From explainable to auditable code — guided by rules, measured by evidence. （不以规矩，不能成方圆）

I work at the intersection of intelligent software engineering and software security, with a current focus on trustworthy code generation. My group studies white-box, auditable generation pipelines that unify correctness × efficiency × security × auditability, embedding decision & evidence into the generation process to move from explainable to auditable code. We also build repo-level analyses and benchmarks for LLM→CodeQL generation/evaluation and study LLM security for code.

Email CV Scholar GitHub LinkedIn

🚀 I’m open to motivated students and collaborators in trustworthy code generation, repo-level security, and LLM security for code.

Research

Trustworthy & Efficient Code Generation. High-performance, low-memory code generation; safety-aware decoding; white-box auditable pipelines (constrained decoding, rollback, candidate ranking, efficiency/security evaluation, evidence chains).
Repo-level Code Security. Vulnerability detection at repository scale; LLM→CodeQL rule generation, repair, and evaluation; joint coverage × quality metrics and benchmarks.
LLM Security for Code. Prompt-injection/jailbreak detection & defense; adversarial examples and data-poisoning robustness; exploit-chain generation & risk assessment; guardrails and red/blue teaming for code LLMs; policy-compliant output filtering.
Deep Architectures & Applications. Transformer / Mamba / KAN optimization and deployment in low-level vision (super-resolution, denoising, dehazing, etc.).

Selected Publications

* corresponding author. A full list is on Google Scholar.

CCF A CCF B CCF C

Chiseling Out Efficiency: Structured Skeleton Supervision for Efficient Code Generation

Yu Yu, Zhihong Sun, Jia Li, Yao Wan, Chuanyi Li, Hongyu Zhang, Ruyun Wang, Tao Huang, Zhi Jin, Ge Li, Chen Lyu*

FSE 2026 CCF A

SemGuard: Real-Time Semantic Evaluator for Correcting LLM-Generated Code

Qinglin Wang, Zhihong Sun, Ruyun Wang, Tao Huang, Zhi Jin, Ge Li, Chen Lyu*

ASE 2025 CCF A

Ensembling Large Language Models for Code Vulnerability Detection: An Empirical Evaluation

Zhihong Sun, Jia Li, Yao Wan, Chuanyi Li, Hongyu Zhang, Zhi Jin, Ge Li, Hong Liu, Chen Lyu*, Songlin Hu

arXiv:2509.12629, 2025

arXiv

Sifting through the Chaff: On Utilizing Execution Feedback for Ranking the Generated Code Candidates

Zhihong Sun, Yao Wan, Jia Li, Hongyu Zhang, Zhi Jin, Ge Li, Chen Lyu*

ASE 2024 CCF A

IRCoCo: Immediate Rewards-Guided Deep Reinforcement Learning for Code Completion

Bolun Li, Zhihong Sun, Tao Huang, Hongyu Zhang, Yao Wan, Ge Li, Zhi Jin, Chen Lyu*

FSE 2024 CCF A

KareCoder: A Knowledge-Enriched Code Generation System

Tao Huang, Zhihong Sun, Zhi Jin, Ge Li, Chen Lyu*

ICSE 2024 CCF A

TACO: Topics in Algorithmic Code Generation Dataset

Rongao Li, Jie Fu, Bowen Zhang, Tao Huang, Zhihong Sun, Chen Lyu*, Guang Liu, Zhi Jin, Ge Li

arXiv:2312.14852, 2023

arXiv

Efficient Dynamic Evolution of Service Composition

Chen Lyu, Wei Jiang, Jiye Wang, Guoliang Lu, Zhiyong Liu, Songlin Hu

IEEE Transactions on Services Computing CCF A, 2018

APISynth: A New Graph-Based API Recommender System

Chen Lyu, Wei Jiang, Yue Liu, Songlin Hu

ICSE 2014 CCF A

News

Mar 2026 EffiSkel accepted to FSE 2026. Congrats to Yuyu!
Jan 2026 Honored to serve on the Program Committee of ASE 2026.
Sep 2025 SemGuard accepted to ASE 2025. Congrats to Qinglin!
Nov 2024 Invited talk at Huawei 2012 Lab (Compiler & Programming Languages LAB, Shanghai): “New explorations in optimizing code recommendation models.”
Nov 2024 Invited talk at ChinaSoft (Intelligent Software Development, Testing & Maintenance Forum): “Complex Code Generation: Strategies for Improving Model Performance.”
Sep 2024 RankEF (code re-ranking) accepted to ASE 2024. Congrats to Zhihong!
Apr 2024 KareCoder accepted to ICSE 2024. Congrats to Tao Huang!
Jan 2024 IRCoCo accepted to FSE 2024. Congrats to Bolun!

Contact

Email: lvchen@sdnu.edu.cn

Google Scholar: OZ_NVPUAAAAJ