Learning Efficient Classifiers Using Genetic Programming

• Main Authors: Jung-Yi Lin, 林忠億
• Other Authors: Been-Chian Chien
• Format: Others
• Language: en_US
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/10474819383528129778

碩士 === 義守大學 === 資訊工程學系 === 90 === Classification is one of the important issues in knowledge discovery and machine learning. An accurate classifier can be applied to many applications. This thesis presents an effective classifier for general data classification based on the learning scheme of genetic programming. The proposed classification learning approach consists of an adaptive incremental learning strategy and a distance-based fitness function for generating the discriminant functions of a classifier from the given samples using genetic programming. In addition, a mechanism called Z-value measure is also developed to resolve the problem of conflict among the discriminant functions. Several well-known datasets are selected from UCI data repository to evaluate the performance of the proposed classifier. The experimental results demonstrate that the proposed classifier is effective in comparison with the previous classifiers.