SEQUENTIAL STOCHASTIC COMBINATORIAL OPTIMIZATION USING HIERARCHICAL REINFORCEMENT LEARNING
Haipeng Chen
This paper introduces a novel hierarchical reinforcement learning framework, wake-sleep option (WS-option), to address sequential stochastic combinatorial optimization problems effectively, demonstrating improved effectiveness, generalizability, and computational efficiency over traditional methods.