Seungjun Oh
Papers
1
Total Citations
2
H-Index
1
About
Seungjun Oh is a researcher advancing the frontiers of Hierarchical Reinforcement Learning (HRL), a paradigm critical for tackling complex, long-horizon tasks with sparse rewards. His work centers on designing intelligent agents that can autonomously decompose high-level goals into manageable subgoals, bridging the gap between abstract planning and low-level control. In his notable 2024 paper, "Novelty-aware Graph Traversal and Expansion for Hierarchical Reinforcement Learning," Oh introduces a method that enhances how high-level policies explore and generate subgoals by leveraging novelty signals during graph-based traversal. This innovation improves sample efficiency and task completion in environments where traditional HRL struggles with exploration. Though early in his career, his contributions are already gaining recognition, with his most-cited work accumulating citations that underscore its relevance to the HRL community. Oh’s research promises to make autonomous systems more adept at navigating complex, real-world scenarios, from robotics to game AI, by enabling more structured and efficient learning hierarchies.
Research Focus
Key Achievements
Top Papers
- 1