Reinforcement Learning in Drug Molecular Design

Reinforcement learning (RL) has emerged as a powerful framework for drug development, offering a natural way to model the iterative and feedback-driven nature of the discovery process. This chapter provides a systematic overview of recent advances in RL-based methods for drug molecular design. We summarize key developments since 2017 and categorize existing approaches according to their underlying molecular representations, including one-dimensional strings, two-dimensional graphs, and three-dimensional geometries. By connecting these methods to their algorithmic foundations and practical applications, we highlight how RL enables efficient exploration of chemical space and goal-directed molecular design. Finally, we discuss current challenges and future opportunities toward more interpretable, data-efficient, and realistic reinforcement learning systems for drug discovery.