Improving gender recognition using genetic algorithms

  • Authors:
  • Abbas Roayaie Ardakany;Sushil J. Louis

  • Affiliations:
  • Dept. of Computer Science and Engineering, University of Nevada, Reno;Dept. of Computer Science and Engineering, University of Nevada, Reno

  • Venue:
  • SEAL'12 Proceedings of the 9th international conference on Simulated Evolution and Learning
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper attacks the problem of gender classification using genetic algorithms. We use local binary patterns and principle component analysis to extract a set of features from face images in the FERET database. The genetic algorithm searches the space of feature subsets to find a set of features that maximizes gender classification accuracy using a support vector machine classifier with hand tuned parameters. Starting with 142 features, the genetic algorithm reduces the feature set size by approximately half resulting in 98.5% accuracy with 100% reliability. This accuracy and reliability are better than when using all 142 features.