Ke Sun

Ke Sun is a Postdoctoral Fellow in Statistics and Computer Science at Harvard University, working with Prof. Susan Murphy. His research develops statistical reinforcement learning methods with applications to healthcare, robotics, and adaptive experimentation.

Sun's work focuses on uncertainty-aware algorithms, distributional RL optimization, and causal inference for adaptive trials. Recent publications address challenges in continual learning, privacy-preserving methods, and robust training against noisy observations.

He has created novel techniques including damped Anderson acceleration for RL convergence and AdaBoost-enhanced graph convolutional networks. Current projects explore optimal treatment allocation strategies for partially observable A/B tests and categorical distributional approaches for exploration.