Feature selection and parameter optimization for support vector machines using particle swarm optimization and harmony search

Jihee Han, Yoonho Seo

Research output: Contribution to journalReview articlepeer-review

Abstract

The present paper proposes a mechanism, Diverse Particle Swarm Optimization and Harmony Search (DPSO_HS), which finds feature subsets and parameter values for Support Vector Machines (SVM) when addressing classification problems by incorporating Particle Swarm Optimization (PSO) and Harmony Search (HS). Specifically, we introduced HS to enhance diversity in the PSO process since it has the advantage of providing diverse solutions as compared to other methodologies, as it considers all solutions in memory when improvising a new solution. For performance evaluation, various datasets with a wide range of features, instances, and classes were considered. DPSO_HS showed an increased diversity and classification accuracy as compared to PSO where statistical significance was found in most datasets. In addition, with two different hybridized approaches based on PSO, we observed that the proposed method showed higher accuracy for most datasets. We also reviewed the results of previous research with identical datasets and found that DPSO_HS achieved higher or equal accuracy rates for most datasets.

Original languageEnglish
JournalInternational Journal of Industrial Engineering : Theory Applications and Practice
Volume28
Issue number1
Publication statusPublished - 2021

Keywords

  • Feature selection
  • Harmony search
  • Meta-heuristics
  • Particle swarm optimization
  • Support vector machines

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering

Fingerprint

Dive into the research topics of 'Feature selection and parameter optimization for support vector machines using particle swarm optimization and harmony search'. Together they form a unique fingerprint.

Cite this