Using genetic algorithms for searching radial basis function centers

Pao Yue-kong Library Electronic Theses Database

Using genetic algorithms for searching radial basis function centers

 

Author: Cho, Kin-wai
Title: Using genetic algorithms for searching radial basis function centers
Degree: M.Sc.
Year: 1997
Subject: Neural networks (Computer science)
Genetic algorithms
Hong Kong Polytechnic University -- Dissertations
Department: Multi-disciplinary Studies
Pages: 63 leaves : ill. ; 30 cm
Language: English
InnoPac Record: http://library.polyu.edu.hk/record=b1250521
URI: http://theses.lib.polyu.edu.hk/handle/200/2822
Abstract: This dissertation proposes to use genetic algorithms to search for the function centers of radial basis function networks. The basis function centers are directly encoded in binary genetic strings. Non-locally tuned centers are found by a genetic algorithm and its performance is evaluated by several experiments such as XOR, overlapped Gaussian distribution clusters and non-overlapped Gaussian distribution clusters. A comparison between the standard RBF using K-means clustering and GA's is performed. Their mean squared error, training time, centers locations and decision surface are studied. The results show that the RBF networks produced by the genetic algorithm give smaller mean squared error, but one of the features of RBF networks is lost. The outputs of the RBF networks trained with the GA may be high in the regions not specified by the training data. This is because non-locally tuned centers are used. In our implementation, the training time is approximately 30 times longer than that of the K-means clustering. A linearly relationship between training time and the number of generations is also observed. It is concluded that the RBF networks produced by the GA is able to achieve good performance (in terms of mean squared error) if sufficient training data is available.

Files in this item

Files Size Format
b12505213.pdf 2.162Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

     

Quick Search

Browse

More Information