Papers
12
Total Citations
1,451
H-Index
8
About
Kaiqing Zhang is a leading researcher in multi-agent reinforcement learning (MARL), distributed optimization, and decision-making theory, whose work sits at the intersection of machine learning, control systems, and game theory. He is perhaps best known for his comprehensive survey, "Multi-Agent Reinforcement Learning: A Selective Overview of Theories and Algorithms," which has accumulated over 1,100 citations and has become a foundational reference for researchers entering the field. His contributions extend beyond surveys: Zhang has made significant theoretical advances in understanding how decentralized agents with networked communication can learn cooperatively, as explored in his work on networked MARL. He has also tackled fundamental questions about the convergence of policy gradient methods, providing rigorous global convergence guarantees that underpin many modern deep RL algorithms. His research on communication-efficient distributed RL addresses practical scalability challenges critical to real-world deployment. More recently, Zhang has explored equilibrium computation in multi-agent settings and the theoretical underpinnings of independent learning in stochastic games. Across his body of work, Zhang consistently bridges mathematical rigor with practical relevance, making him an influential voice shaping both the theory and application of modern reinforcement learning.
Research Focus
Key Achievements
Top Papers
- 1Multi-Agent Reinforcement Learning: A Selective Overview of Theories and Algorithms1,121 citations · 2021
- 2
- 3
- 4
- 5Communication-Efficient Distributed Reinforcement Learning41 citations · 2018
- 6
- 7Independent learning in stochastic games9 citations · 2023
- 8
- 9
- 10