KSATS-HH: A Simulated Annealing Hyper-Heuristic with Reinforcement Learning and Tabu-Search

The term hyper-heuristics describes a broad range of techniques that attempt to provide solutions to an array of problems from different domains. This paper describes the implementation of a hyperheuristic developed for entry into the Cross-domain Heuristic Search Challenge organised by the Automated Scheduling, Optimisation and Planning research group at the University of Nottingham. An algorithm taking inspiration from the Simulated Annealing, Tabu Search and Reinforcement Learning paradigms was implemented and incorporated into the supplied Hyper-heuristics Flexible framework with initial results proving promising, outperforming the benchmark results given.