Michael Wan
Papers
1
Total Citations
5
H-Index
1
About
Michael Wan is a researcher advancing the frontiers of deep reinforcement learning (RL), with a particular focus on sample efficiency and knowledge transfer across complex decision-making tasks. His most-cited work, "Mutual Information Based Knowledge Transfer Under State-Action Dimension Mismatch" (2020, 5 citations), tackles a critical bottleneck in RL: the high sample complexity that plagues many algorithms when learning from scratch. By introducing a mutual information framework, Wan enables effective transfer of learned policies even when source and target tasks have mismatched state-action spaces—a common yet underexplored challenge. This contribution addresses fundamental issues like credit assignment and reward sparsity, offering a path toward more scalable and data-efficient RL systems. While his citation count is modest, his work reflects a deep engagement with core theoretical problems that underpin practical RL deployment. Wan’s research is particularly relevant for students and practitioners seeking to understand how intelligent agents can leverage prior knowledge to accelerate learning in novel environments, making him a thoughtful voice in the ongoing effort to bridge the gap between RL theory and real-world application.
Research Focus
Key Achievements
Top Papers
- 1