Liuyuan Wen
I am a Ph.D. student at the School of Intelligence Science and Technology, Nanjing University (NJU), advised by Prof. Wenbin Li. My research focuses on understanding and improving intelligent systems through large language models, mechanistic interpretability, and brain-inspired artificial intelligence.
Before transitioning into artificial intelligence, I studied physics at Tianjin University (TJU) and the University of Science and Technology of China (USTC), receiving my B.S. and M.S. in Physics, respectively. This interdisciplinary background continues to shape how I approach representation, reasoning, and scientific discovery in modern AI systems.
Research Interests
Large Language Models
Internal representations, reasoning behavior, and reliable computation in foundation models.
Mechanistic Interpretability
Geometric and causal analyses of how learned representations support model behavior.
Brain-Inspired AI
Interpretable learning systems inspired by cognition, neuroscience, and structured reasoning.
Selected Publications
Negative Adjustment for Contrastive Learning in Audio-Visual Generalized Zero-Shot Learning
Liuyuan Wen
International Conference on Neural Information Processing (ICONIP), 2024.
Education
School of Intelligence Science and Technology, Nanjing University
University of Science and Technology of China
Tianjin University
Recent Work & Recognition
LLM-based ECA Rule Mining
Student lead for an industry-sponsored research project developing a multi-agent framework for Event–Condition–Action rule mining with domain-adapted large language models.
3rd World Science Intelligence Competition
Third Prize, Materials Design Track, for designing and implementing a diffusion-based 3D molecular generation method in a unified latent space.