Zhang-Wei Hong

Research: I develop reinforcement learning (RL) methods for computational discovery—finding novel solutions in domains ranging from materials science to robotics. My work addresses two fundamental challenges in applying RL to real-world discovery:

• Learning from sparse feedback: Discovery problems often provide limited reward signals, making it difficult for RL agents to learn effectively. My research develops principled approaches to accelerate learning despite sparse supervision. [NeurIPS'24, NeurIPS'23, ICML'23, ICML'23, ICLR'23, ICLR'22, ICLR'22]
• Generating diverse solutions: Many discovery problems require multiple high-quality candidates rather than a single optimum. I develop methods that produce diverse solution sets for applications like drug discovery and AI safety. [ICLR'24]

Bio: I am a Principal Investigator and Research Staff Member at the MIT-IBM Watson AI Lab. I received my Ph.D. in Electrical Engineering and Computer Science from MIT, advised by Prof. Pulkit Agrawal. I was a recipient of the Qualcomm Innovation Fellowship (2024). Prior to MIT, I earned my B.S. and M.S. from National Tsing Hua University, where I worked with Prof. Chun-Yi Lee and Prof. Min Sun. I have also collaborated with Prof. Jan Peters at TU Darmstadt and conducted research at Preferred Networks.

template from jonbarron