Le Tien Dung
Papers
1
Total Citations
6
H-Index
1
About
Le Tien Dung is a researcher focused on advancing reinforcement learning, particularly in complex, non-Markovian environments where traditional Markov assumptions break down. His most-cited work, "Efficient experience reuse in non-Markovian environments" (2008, 6 citations), tackles the critical challenge of reducing learning time when recurrent neural networks are employed to predict Q-values. Dung’s key contribution lies in developing a novel method for experience reuse, enabling agents to more efficiently leverage past interactions to accelerate learning in partially observable settings. This work addresses a fundamental bottleneck in reinforcement learning—the high computational cost of training recurrent architectures—and offers practical strategies for improving sample efficiency. While his citation count is modest, Dung’s research holds significance for students and practitioners working on real-world applications where environments are inherently non-Markovian, such as robotics, game AI, and autonomous systems. His focus on experience reuse provides a foundation for scalable, memory-efficient learning, making his contributions a valuable reference for those seeking to enhance agent performance in dynamic, uncertain domains.
Research Focus
Key Achievements
Top Papers
- 1Efficient experience reuse in non-Markovian environments6 citations · 2008