Xueguang Lyu

Xueguang Lyu

CS Ph.D. Candidate

Northeastern University

Biography

Xueguang Lyu (吕雪广) is a CS Ph.D. student at Khoury College of Computer Sciences of Northeastern University, advised by Christopher Amato. Xueguang’s research interests include reinforcement learning, artificial intelligence, and multi-agent system.

Xueguang is interested in enabling effective multi-agent exploration and cooperation in game-theoritic simulations as well as in real-world tasks. His recent work focuses on analyzing multi-agent actor-critic methods in centralized training settings, and utilizing centralized training to learn decentralized cooperating policies.

Interests
  • Reinforcement Learning
  • Mulit-agent System
Education
  • PhD in Computer Science, in progress

    Northeastern University

  • MS in Computer Science, 2018

    Northeastern University

  • BS in Informatics, 2016

    University of California, Irvine

Publications

On Centralized Critics in Multi-Agent Reinforcement Learning. In JAIR, 2023.
A Deeper Understanding of State-Based Critics in Multi-Agent Reinforcement Learning. In AAAI, 2022.
Local Advantage Actor-Critic for Robust Multi-Agent Deep Reinforcement Learning. In MRS, Best Paper Award Finalist, 2021.
Contrasting Centralized and Decentralized Critics in Multi-Agent Reinforcement Learning. In AAMAS, Best Paper Award Finalist, 2021.