Machine Learning & Mechanism Design

Advancing the intersection of artificial intelligence, continuous learning, and intelligent physical systems.

Explore Our Research

About Us

Welcome to the MLMD (Machine Learning & Mechanism Design) Group at the School of Artificial Intelligence, Jilin University (JLU), led by Dr. Changyi Ma. We investigate robust and adaptive machine learning methods for complex intelligent systems, with the goal of bridging fundamental AI research and real-world engineering and design problems. Our research spans Continual Learning, Information Retrieval, AI-driven Optimization and Mechanism Design, LLMs and Intelligent Agents.

Current Students

PhD Students

  • 26 Fall Mr. Zhiqiang Guo (MLLM)
  • 25 Fall Mr. Yicheng Zhao (LLM Agents), Mr. Bicong Sang (Evolutionary Computing)

Master Students

  • 26 Fall Mr. Bingyu Cao (Continual Learning), Mr. Zishan Zhao (LLM Agents), Mr. Fulin Min, Mr. Wenjun Chen
  • 25 Fall Mr. Yingying Wu (LLM Agents), Mr. Kaiquan Hu (LLM Agents), Mr. Liangenrong Shi (AI4Sci), Mr. Hanran Zhou (LLM Agents)

Undergraduate Students

  • 25 Fall Mr. Zijie Xi (AI4PDE), Mr. Yibo Wang (AI4PDE)
  • 24 Fall Miss. Chang Wang (LLM Agents & Security), Mr. Ruoyu Zhang (LLM Agents & Security), Mr. Huiyu Gong (Continual Learning)

Research Projects

Construction of a Lithium Battery Knowledge Base Based on LLM

A comprehensive RAG-based framework for automatic scientific literature processing and intelligent question-answering in the lithium battery domain.

View Details →

Selected Publications

A Low-Rank Perspective on Similarity Matrix Completion (TKDE 2026, CCF A)

Changyi Ma, Runsheng Yu, Xiao Chen, Youzhi Zhang, Zhen Lei.

Deep Noisy-Label Learning via Effective Rank Reduction (ECCV 2026, CCF B)

Changyi Ma, Zihan Fang, Lihua Zhou, Tao Li, Wenyu Liu, Runsheng Yu, Xuan Song.

Diagnosing Parametric Contamination in LoRA-Based RAG Distillation (CIKM 2026, CCF B)

Yingying Wu, Xiao Chen, Changyi Ma, Xuan Song.

Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training (ICML 2026, CCF A)

Song Lai, Haohan Zhao, Rong Feng, Changyi Ma, Wenzhuo Liu, Hongbo Zhao, Xi Lin, Dong Yi, Min Xie, Qingfu Zhang, Hongbin Liu, Gaofeng Meng, Fei Zhu.

Lookahead-GCG: Improving Universal Multi-Model Optimization-Based Jailbreaking Attacks via Stochastic Nesterov Optimization. (ICML 2026, CCF A)

Rong Feng, Haohan Zhao, Shiqin Tang, Geng Liu, Song Lai, Meng Wang, Shuxin Zhuang, Yuqi Zha, Changyi Ma, Runsheng Yu

Gradient-Guided Epsilon Constraint Method for Online Continual Learning. (NeurIPS 2025, CCF A)

Song Lai, Changyi Ma, Fei Zhu, Zhe Zhao, Xi Lin, Gaofeng Meng, Qingfu Zhang.