Liansheng Zhuang

University of Science and Technology of China

Papers

Total Citations

H-Index

About

Liansheng Zhuang is a prominent researcher in the field of Deep Reinforcement Learning (DRL), with a focused expertise in addressing the challenge of efficient learning in environments with sparse rewards. His major contributions center on advancing Hindsight Experience Replay (HER), a critical technique for continuous control tasks like robotic arm manipulation. Zhuang introduced "Soft Hindsight Experience Replay" (2020, 11 citations), which enhances HER's robustness and learning efficiency by softening the hindsight transitions, reducing brittleness in complex DRL settings. He further refined this approach with "Quantile Regression Hindsight Experience Replay" (2020, 3 citations), integrating quantile regression to improve value estimation and policy stability. These works have been instrumental in making DRL more practical for real-world robotic applications, where reward signals are often scarce. Zhuang's research demonstrates a clear trajectory of innovation in reinforcement learning, offering foundational improvements that enable more reliable and adaptive autonomous systems. His work is particularly notable for its direct impact on continuous control problems, bridging the gap between theoretical DRL advances and tangible robotic performance.