A Parallel Classification Algorithm Based on Hybrid Genetic Algorithm

In this paper, a parallel classification algorithm based on an improved hybrid genetic algorithm (PC-HGA) is presented. It attempts to solve the problems of lower classification rule quality, more redundancy rules after optimizing generations and classification accuracy using the traditional genetic algorithm in classification mining. A rule extraction approach to improve the classification accuracy and condense the classification rule set is also given. In order to further improve the efficiency of classification mining, the master-slave parallel computing mode is adopted in PC-HGA. Experiments of PC-HGA algorithm are carried out on two benchmark datasets: iris and dermatology from UCI machine-learning repository. The experimental results show that PC-HGA has good speedup performance and can discover a set of the succinct, efficient and understandable classification rules