Jiangpeng Yan
Papers
2
Total Citations
5
H-Index
2
About
Jiangpeng Yan is a researcher advancing the frontiers of multi-goal reinforcement learning (RL), with a focus on overcoming fundamental challenges in planning and robotic manipulation. His work directly addresses two persistent hurdles in the field: sparse rewards and sample inefficiency. Yan’s key contribution lies in refining the Hindsight Experience Replay (HER) framework, a popular method that relabels goals to learn from failed attempts. In his most-cited papers, including “Multi-Step Hindsight Experience Replay with Bias Reduction for Efficient Multi-Goal Reinforcement Learning” (2023, 3 citations) and its precursor from 2021 (2 citations), he identifies and mitigates inherent biases in multi-step HER. By introducing bias-reduced relabeling strategies, Yan’s approach enables RL agents to learn more efficiently from limited interaction data, making them more practical for complex, real-world tasks. Though his citation counts are modest, his theoretical contributions are notable for their precision in addressing a known weakness in a widely used algorithm. Yan’s work is particularly relevant for students and researchers seeking to improve sample efficiency in goal-conditioned RL, offering a principled path toward more robust and scalable learning systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2