Affordable Access

Access to the full text

A Novel Adaptive Kernel for the RBF Neural Networks

Authors
  • Khan, Shujaat1
  • Naseem, Imran2, 3
  • Togneri, Roberto3
  • Bennamoun, Mohammed4
  • 1 Iqra University, Defence View, Faculty of Engineering Science and Technology, Shaheed-e-Millat Road (Ext.), Karachi, 75500, Pakistan , Karachi (Pakistan)
  • 2 Karachi Institute of Economics and Technology, College of Engineering, Korangi Creek, Karachi, 75190, Pakistan , Karachi (Pakistan)
  • 3 The University of Western Australia, School of Electrical, Electronic and Computer Engineering, 35 Stirling Highway, Crawley, WA, 6009, Australia , Crawley (Australia)
  • 4 The University of Western Australia, School of Computer Science and Software Engineering, 35 Stirling Highway, Crawley, WA, 6009, Australia , Crawley (Australia)
Type
Published Article
Journal
Circuits, Systems, and Signal Processing
Publisher
Springer US
Publication Date
Jul 30, 2016
Volume
36
Issue
4
Pages
1639–1653
Identifiers
DOI: 10.1007/s00034-016-0375-7
Source
Springer Nature
Keywords
License
Yellow

Abstract

In this paper, we propose a novel adaptive kernel for the radial basis function neural networks. The proposed kernel adaptively fuses the Euclidean and cosine distance measures to exploit the reciprocating properties of the two. The proposed framework dynamically adapts the weights of the participating kernels using the gradient descent method, thereby alleviating the need for predetermined weights. The proposed method is shown to outperform the manual fusion of the kernels on three major problems of estimation, namely nonlinear system identification, patter classification and function approximation.

Report this publication

Statistics

Seen <100 times