A novel classifier with the immune-training based wavelet neural network

  • Authors:

  • Lei Wang;Yinling Nie;Weike Nie;Licheng Jiao

  • Affiliations:

  • School of Computer Science and Engineering, Xi'an University of Technology, Xi'an, China;School of Computer Science and Engineering, Xi'an University of Technology, Xi'an, China;School of Electronic Engineering, Xidian University, Xi'an, China;School of Electronic Engineering, Xidian University, Xi'an, China

  • Venue:
  • ISNN'05 Proceedings of the Second international conference on Advances in neural networks - Volume Part II
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

After analyzing the classification and training ability of a wavelet neural network (WNN), a novel WNN learning scheme integrating immunity based evolutionary algorithm (IDEA) is proposed, in which, IDEA is an evolutionary algorithm with an embedded immune mechanism. When WNN is used as a classifier, the process of seeking the least mean-square error (LMS) of an optimal problem is equivalent to that of finding the wavelet feature with maximal separability, namely, maximizing its separable division. On the other hand, with the capability of robust learning of its evolutionary process, IDEA is able to eliminate local degenerative phenomenon due to blindfold behaviors of original operators in the existing evolutionary algorithms. In the case of the twin-spiral problem, experimental simulation shows the feasibility of WNN training with the IDEA based learning algorithm.