Sili Huang

Papers

Total Citations

H-Index

About

Sili Huang is a researcher advancing the frontiers of reinforcement learning (RL), with a focus on generalization, continual learning, and multi-agent systems. Their work tackles fundamental challenges in making RL agents robust, adaptable, and effective in dynamic environments. Huang’s most cited paper, “Learning Generalizable Agents via Saliency-Guided Features Decorrelation” (2023, 3 citations), addresses the critical problem of visual-based RL agents failing to generalize to unseen environmental variations, such as background noise or changes in task-relevant features. By introducing saliency-guided decorrelation, this work enhances agent robustness, a key step toward real-world deployment. In “Continual Diffuser (CoD): Mastering Continual Offline RL With Experience Rehearsal” (2025, 2 citations), Huang pioneers the use of diffusion models for continual offline RL, enabling agents to learn sequentially without catastrophic forgetting—a vital capability for lifelong robotic learning. Their earlier work, “Distributional Reward Estimation for Effective Multi-Agent Deep Reinforcement Learning” (2022, 2 citations), tackles reward uncertainty in multi-agent settings, improving policy learning for applications like robotics and autonomous driving. With a growing citation impact and a focus on practical, scalable solutions, Sili Huang is shaping the next generation of intelligent, adaptive agents.