Asian Journal of Information Technology

Year: 2006
Volume: 5
Issue: 4
Page No. 380 - 387

Supervised Machine Learning by Generation of Rules: Optimization of the Size of the Base of Rules of Training by the Method of Inclusion

Authors : M. Kaddeche and N. Doghmane

Abstract: Within the framework of supervised induction, in this study, a method to minimize the number of rules for a training base is proposed based on optimisation criterion of the simultaneous functions. This method consists in determining the redundant terms (included) in order to lead to a non redundant base. The method is used in various fields such as the automatism to minimise the realisation costs of the simultaneous functions. By analogy, we try through this study, to apply this method to various basic examples in order to remove the redundant rules of a training base and to lead to a base made only prime rules. This method is based on the comparison of two of the same rules classifies. Each rule is made of two parts: (1) The membership area of attributes. (2) The degree of belief of these rules. For the first part, the inclusion notion is applied. However, for the second rule which is it is a number, the superiority or inferiority is used. We present here the experimental tests with the IRIS data base. The obtained principal results and their comparison with other method are given. These results are satisfactory and constitute an additional validation of our method.

How to cite this article:

M. Kaddeche and N. Doghmane , 2006. Supervised Machine Learning by Generation of Rules: Optimization of the Size of the Base of Rules of Training by the Method of Inclusion. Asian Journal of Information Technology, 5: 380-387.

Design and power by Medwell Web Development Team. © Medwell Publishing 2022 All Rights Reserved