Optimal classifier design method using hierarchical fair competition model based parallel genetic algorithm

Heesung Lee, Sungjun Hong, Euntai Kim

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

By the appropriate editing of the reference set and the judicious selection of features, we can obtain optimal classifier which maximizes the classification accuracy while saving computational time and memory resources. In this paper, a new simultaneous reference set editing and feature selection for an optimal classifier is proposed. Genetic algorithm (GA) based simultaneous editing of the reference set and feature selection to design optimal classifier is receiving attention. However, the problem to find an optimal classifier has very large search spaces. Compared with the simple genetic algorithm (SGA), the hierarchical fair competition parallel genetic algorithm (HFC-PGA) exhibits a promising performance when dealing with huge search spaces, high-dimensionality, and multimodality of the search problems. Therefore, we develop a design methodology for optimal classifier, which deals with simultaneous reference set editing and feature selection using HFC-PGA. Experiments are performed with UCI machine learning repository to show the performance of the proposed algorithm.

Original languageEnglish
Title of host publicationICCAS-SICE 2009 - ICROS-SICE International Joint Conference 2009, Proceedings
Pages2907-2910
Number of pages4
Publication statusPublished - 2009
EventICROS-SICE International Joint Conference 2009, ICCAS-SICE 2009 - Fukuoka, Japan
Duration: 2009 Aug 182009 Aug 21

Publication series

NameICCAS-SICE 2009 - ICROS-SICE International Joint Conference 2009, Proceedings

Other

OtherICROS-SICE International Joint Conference 2009, ICCAS-SICE 2009
CountryJapan
CityFukuoka
Period09/8/1809/8/21

All Science Journal Classification (ASJC) codes

  • Information Systems
  • Control and Systems Engineering
  • Industrial and Manufacturing Engineering

Fingerprint Dive into the research topics of 'Optimal classifier design method using hierarchical fair competition model based parallel genetic algorithm'. Together they form a unique fingerprint.

Cite this