Affiliation:
1. Department of Informatics and Computers, University of Ostrava, 30. Dubna 22, 70103 Ostrava, Czech Republic
Abstract
The problem of optimisation methods is the stagnation of population P, which results in a local solution for the task. This problem can be solved by employing an archive for good historical solutions outperformed by the new better offspring. The archive A was introduced with the variant of adaptive differential evolution (DE), and it was successfully applied in many adaptive DE variants including the efficient jSO algorithm. In the original jSO, the historical good individuals replace the random existing positions in A. It causes that outperformed historical solution from P with lower quality to replace the stored solution in A with better quality. In this paper, a new approach to replace individuals in archive A more progressively is proposed. Outperformed individuals from P replace solutions in the worse part of A based on the function value. The portion of A selected for replacement is controlled by the input parameter, and its setting is studied in this experiment. The proposed progressive archive is employed in the original jSO. Moreover, the Eigenvector transformation of the individuals for crossover is applied to increase the efficiency for the rotated optimisation problems. The efficiency of the proposed progressive archive and the Eigen crossover are evaluated using the set of 29 optimisation problems for CEC 2024 and various dimensionality. All the experiments were performed on a standard PC, and the results were compared using the standard statistical methods. The newly proposed algorithm with the progressive archive approach performs substantially better than the original jSO, especially when 20 or 40% of the worse individuals of A are set for replacement. The Eigen crossover increases the performance of the proposed jSO algorithm with the progressive archive approach. The estimated time complexity illustrates the low computational demands of the proposed archive approach.
Reference20 articles.
1. Storn, R., and Price, K.V. (1995). Differential Evolution—A Simple and Efficient Adaptive Scheme for Global Optimization over Continuous Spaces, International Computer Science Institute. Available online: https://cse.engineering.nyu.edu/~mleung/CS909/s04/Storn95-012.pdf.
2. Differential evolution—A Simple and Efficient Heuristic for Global Optimization over Continuous Spaces;Storn;J. Glob. Optim.,1997
3. Recent advances in differential evolution—An updated survey;Das;Swarm Evol. Comput.,2016
4. Differential Evolution: A Survey of the State-of-the-Art;Das;IEEE Trans. Evol. Comput.,2011
5. No Free Lunch Theorems for Optimization;Wolpert;IEEE Trans. Evol. Comput.,1997