Solving Continuous Optimization Problems with a New Hyperheuristic Framework

Continuous optimization is a central task in computer science. Hyperheuristics prove to be an effective mechanism for intelligent operator selection and generation for optimization problems. In this paper we propose a two level hyperheuristic framework for continuous optimization problems. The base level is used to optimize the problem with operator sequences that are modeled by a nested Markov chain, while the hyper level searches the operator sequence and parameter space with simulated annealing. The experimental results show that the proposed approach matches the performance of another state-of-the-art hyperheuristic using significantly less operators and computational time. The model outperforms the simple metaheuristic operator approach and the random hyperheuristic search strategy.