Wrapper approach for learning neural network ensemble by feature selection

  • Authors:

  • Haixia Chen;Senmiao Yuan;Kai Jiang

  • Affiliations:

  • College of Computer Science and Technology, Jilin University, Changchun, China;College of Computer Science and Technology, Jilin University, Changchun, China;The 45th Research Institute of CETC, Beijing, China

  • Venue:
  • ISNN'05 Proceedings of the Second international conference on Advances in Neural Networks - Volume Part I
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

A new algorithm for learning neural network ensemble is introduced in this paper. The proposed algorithm, called NNEFS, exploits the synergistic power of neural network ensemble and feature subset selection to fully exploit the information encoded in the original dataset. All the neural network components in the ensemble are trained with feature subsets selected from the total number of available features by wrapper approach. Classification for a given intance is decided by weighted majority votes of all available components in the ensemble. Experiments on two UCI datasets show the superiority of the algorithm to other two state of art algorithms. In addition, the induced neural network ensemble has more consistent performance for incomplete datasets, without any assumption of the missing mechanism.