AN EFFICIENT HIERARCHICAL FUZZY ENTROPY CLASSIFIER TECHNIQUE FOR FEATURE SUBSET SELECTION

Feature selection is an important preprocessing technique in classifying systems. A set of attributes which are relevant, irrelevant or redundant is used and from the viewpoint of managing a dataset which can be huge, reducing the number of attributes by selecting only the relevant ones is desirable. In doing so, a higher performance with lower computational ef ort is expected. First, numeric features are discredited to construct the membership function of each fuzzy set of a feature. Then, the feature subset is selected based on the fuzzy entropy measure focusing on boundary samples. This method can select relevant features to obtain higher average classification accuracy rates. It is found that the existing method have some limitations in selection of correct features. In this paper, a new method is proposed for dealing with feature subset selection based on Hierarchical fuzzy entropy measures (FC-HFE) for handling classification problems. The proposed method is designed on the basis of the fuzzy entropy measures. The proposed algorithms can be applied to several classification problems. The proposed FC-HFE can improve the classification accuracy and overcome some of the drawbacks in the existing method. Finally, the FC-HFE is applied to evaluate the classification performance for iris and spiral databases. The simulation results illustrate that the classification rate of the proposed FC-HFE is better than earlier methods.