Marcelo Li Koga
Papers
2
Total Citations
47
H-Index
2
About
Marcelo Li Koga is a researcher whose work lies at the intersection of reinforcement learning and robotics, with a particular focus on knowledge transfer and risk-aware decision-making. His key contribution is the development of **stochastic abstract policies**—a framework that allows agents to generalize knowledge across tasks, significantly accelerating learning in complex environments. In his most-cited paper, "Stochastic Abstract Policies: Generalizing Knowledge to Improve Reinforcement Learning" (2014, 45 citations), Koga demonstrated how abstract policies can capture reusable behavioral patterns, enabling agents to avoid the costly process of learning from scratch. This work has been foundational for researchers seeking to improve sample efficiency in RL. Koga extended these ideas to robotics in "Reusing Risk-Aware Stochastic Abstract Policies in Robotic Navigation Learning" (2014), where he introduced mechanisms to incorporate risk awareness, ensuring safer and more robust navigation. While his citation count reflects a focused but impactful body of work, Koga’s contributions are notable for bridging theoretical RL advances with practical robotic applications, offering a pathway toward more intelligent and efficient autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2