A deep reinforcement learning based hyper-heuristic for combinatorial optimisation with uncertainties

Yuchang Zhang, Ruibin Bai, Rong Qu, Chaofan Tu, Jiahuan Jin

Abstract: Highlights•A hyper-heuristic framework is proposed for online combinatorial optimisation.•A deep reinforcement learning agent is first used in hyper-heuristic.•The method holds better convergence performance than the existing methods.•Our approach improves the interpretability of traditional DRL.