Dijun Luo
Papers
5
Total Citations
48
H-Index
3
About
Dijun Luo is a researcher specializing in deep reinforcement learning, with particular expertise in cooperative multi-agent systems, model-based learning, and goal-conditioned reinforcement learning. His work addresses some of the field's most pressing challenges: enabling agents to operate effectively in complex, real-world environments where action spaces, sample efficiency, and reward sparsity present significant obstacles. Luo's most influential contribution, "Structured Cooperative Reinforcement Learning With Time-Varying Composite Action Space" (2021, 26 citations), tackles the underexplored problem of composite and time-varying action spaces in practical multi-agent settings — a critical step toward deploying reinforcement learning beyond controlled game environments. His work on model-ensemble exploration and exploitation (2021, 15 citations) advances sample efficiency in model-based deep reinforcement learning by improving uncertainty estimation and exploration planning, areas vital for robotics applications. Luo has also made notable strides in multi-goal reinforcement learning, developing bias-reduction techniques for Hindsight Experience Replay that address longstanding issues of sparse rewards and sample inefficiency in robot manipulation tasks. Across his body of work, Luo consistently bridges theoretical rigor with practical applicability, making his research particularly valuable for robotics, autonomous systems, and real-world AI deployment.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5