Matrix-based Genetic Algorithm with Principal Component Analysis based Adaptive Parameter Control

YisuGe, Zhou Zhou, Weipeng Tang, Jun Zhang

Keywords: Matrix Genetic Algorithm, Principal Component Analysis, Adaptive Parameter Control

Abstract: As a classical method for solving complex problems, genetic algorithms (GA) play a significant role in addressing practical issues. However, as the complexity of problems increases, the large number of model variables poses significant challenges for traditional genetic algorithms, including heavy computational burdens and prolonged solving times. Effectively enhancing the problem-solving capabilities of genetic algorithms has become one of the important research topics in the field of artificial intelligence. To this end, this paper focuses on the parameter optimization problem in genetic algorithms utilizing matrix calculations, and proposes a matrix-based genetic algorithm with adaptive parameter control by principal component analysis. To improve its convergence speed in solving large-scale complex problems, the distribution of individuals within the population matrix is compressed and analyzed using principal component analysis, thereby adaptively adjusting the crossover and mutate parameters of genetic algorithm. For applying the matrix-based genetic algorithm on the broader range of problems, the framework is expanded with floating-point encoding. Additionally, to validate the performance of the proposed algorithm, it is compared with the original matrix-based genetic algorithm. The experimental results demonstrate that the new algorithm effectively enhances both the performance and stability of the algorithm.

Submission Number: 21