New Hybridization Algorithm of Differential Evolution and Particle Swarm Optimization for Efficient Feature Selection

  • Meng Ang Koon
    Faculty of Engineering, Technology and Built Environment, UCSI University
  • Bin Mohamed Juhari Mohd Rizon
    Faculty of Engineering, Technology and Built Environment, UCSI University
  • Hong Lim Wei
    Faculty of Engineering, Technology and Built Environment, UCSI University
  • Sun Tiang Sew
    Faculty of Engineering, Technology and Built Environment, UCSI University
  • Kit Ang Chun
    Faculty of Engineering, Technology and Built Environment, UCSI University
  • Eiyda Hussin Eryana
    Faculty of Engineering, Technology and Built Environment, UCSI University
  • Pan Li
    Faculty of Engineering, Technology and Built Environment, UCSI University
  • Hui Chong Ting
    Faculty of Engineering, Technology and Built Environment, UCSI University

Description

Feature selection is a popular pre-processing technique applied to enhance the learning performances of machine learning models by removing irrelevant features without compromising their accuracies. The rapid growth of input features in big data era has increased the complexities of feature selection problems tremendously. Given their excellent global search ability, differential evolution (DE) and particle swarm optimization (PSO) are considered as the promising techniques used to solve feature selection problems. In this paper, a new hybrid algorithm is proposed to solve feature selection problems more effectively by leveraging the strengths of both DE and PSO. The proposed feature selection algorithm is reported to achieve an average accuracy of 89.03% when solving 7 datasets obtained from UCI Machine Learning Repository.

Journal

Details 詳細情報について

  • CRID
    1390854717509218304
  • DOI
    10.5954/icarob.2022.os22-1
  • ISSN
    21887829
  • Text Lang
    en
  • Data Source
    • JaLC
    • Crossref
  • Abstract License Flag
    Disallowed

Report a problem

Back to top