Quantum-Inspired Evolution Strategy

摘要

Evolution strategy is a suitable method for solving numerical optimization problems whose main characteristic is self adaption of the mutation step size. Finding the promising region in the search space is beneficial in optimization problems. However, in the contemporary ES the next generation is produced in a hyper ellipse and the direction to the optimum is not determined correctly. Therefore it is possible that the mutants are produced in unpromising regions which leads to unsatisfactory convergence. To alleviate this deficiency a novel evolution strategy which is inspired by the quantum computing is proposed in this paper. The proposed algorithm which is called quantum-inspired evolution strategy (QES) can improve the convergence speed and the accuracy by modifying the mutation direction. To demonstrate the effectiveness and applicability of the proposed method, several experiments on a set of numerical optimization problems are carried out. The results show that QES is superior to conventional ES in terms of convergence speed, accuracy and robustness.