Longxiang Shi

Zhejiang University of Science and Technology

Papers

Total Citations

H-Index

About

Longxiang Shi is a researcher at the forefront of reinforcement learning (RL), with a primary focus on advancing the stability and efficiency of evolution strategies (ES) in complex decision-making tasks. His most-cited work, "Maximum Entropy Reinforcement Learning with Evolution Strategies" (2020), tackles a critical bottleneck in ES-based RL methods—their notorious instability. By integrating maximum entropy principles, Shi introduces a novel framework that balances exploration and exploitation, significantly improving convergence reliability without sacrificing the low computational costs and high scalability that make ES attractive. This contribution has garnered 6 citations, reflecting its growing influence among researchers seeking robust alternatives to gradient-based RL. Shi’s work bridges the gap between evolutionary optimization and modern RL, offering practical solutions for challenging environments where traditional methods falter. His research is particularly valuable for students and practitioners aiming to deploy scalable, stable RL systems in robotics, game playing, or real-world control tasks. Through this key paper, Shi has established himself as a thoughtful innovator in the ongoing quest to make ES a viable, high-performance tool for reinforcement learning.