|
This is Enpei Zhang, a second-year Computer Science Ph.D. student at Dartmouth College, where I am fortunate to be advised by Prof. Yujun Yan. Previously, I earned my B.S. in Information Engineering from Shanghai Jiao Tong University, where I worked with Prof. Siheng Chen at Cooperative Medianet Innovation Center (CMIC). / / / CV / Google Scholar |
|
|
My research interests center on AI agents and multi-agent systems, with an emphasis on how agent interactions and workflow structures shape performance, efficiency, and reliability. Recently, I have focused on principled methods for automatic agent workflow orchestration, grounded in game theory and scheduling theory to improve efficiency and incentive alignment. |
|
|
|
Enpei Zhang, Jingyi Chai, Rui Ye, Yanfeng Wang, Siheng Chen Nature Communications, 2025 Paper Proposed iPFL, an incentive-compatible personalized federated learning framework that formulates federated training as a multi-agent market, enabling heterogeneous data holders to collaboratively train personalized models with formal guarantees of individual rationality and incentive compatibility, while achieving strong performance and economic utility across diverse training tasks including simple classification and LLM SFT. Multi-Agent System Game Theory Federated Learning |
|
Chiyu Ma, Enpei Zhang, Yilun Zhao, Wenjun Liu, Yaning Jia, Peijun Qing, Lin Shi, Arman Cohan, Yujun Yan, Soroush Vosoughi EMNLP 2025 Paper Conducted a systematic evaluation of bias behaviors in multi-agent LLM-as-Judge frameworks, modeling evaluation as an interactive multi-agent system rather than a single-judge process. The results reveal the counterintuitive finding that multi-agent debate can exacerbate, rather than reduce, four distinct biases over multiple debate rounds. Multi-Agent System LLM Bias |
|
Zheng Huang, Enpei Zhang, Yinghao Cai, Weikang Qiu, Carl Yang, Elynn Chen, Xiang Zhang, Rex Ying, Dawei Zhou, Yujun Yan ICLR 2026 Paper Demonstrates that fMRI signals align better with structured text representations than vision model representation; and proposes PRISM, a text-based framework that improves fmri-to-image reconstruction via object-centric diffusion. AI for Science Representation Learning |
|
Graduate Teaching Assistant
|
|
Reviewer:
|
|
|
Design and source code from Jon Barron's website.
|