Feature Weights Determining of Pattern Classification by Using a Rough Genetic Algorithm with Fuzzy Similarity Measure

摘要

The classification problem is one of the typical problems encountered in data mining and machine learning. In this paper, a rough genetic algorithm (RGA) is applied to the classification problem in an undetermined environment based on a fuzzy distance function by calculating attribute weights. The RGA, a genetic algorithm based on rough values, can complement the existing tools developed in rough computing. Computational experiments are conducted on benchmark problems downloaded from UCI machine learning databases. Experimental results, compared with the usual GA and C4.5 algorithms, verify the efficiency of the developed algorithm. Furthermore, the weights acquired by the proposed learning method are applicable not only to fuzzy similarity functions but also to any similarity functions. As an application, a new distance metric called weighted discretized value difference metric (WDVDM) is proposed. Experimental results show that WDVDM is an improvement on the discretized value difference metric (DVDM).