A Feature Subset Evaluation Method Based on Multi-objective Optimization

摘要: To remove the irrelevant and redundant features from high-dimensional data while ensuring classification accuracy, a supervised feature subset evaluation method based on multi-objective optimization has been proposed in this paper. Four aspects, sparsity of space, information loss degree stability, were took into account Multi-objective functions constructed. Then popular NSGA-II algorithm was used for four objectives selection process. Finally selected obtained weight vector according criteria. The tested 4 standard sets using two kinds classifier. experiment results show that can guarantee higher accuracy even though only few numbers than other methods. On hand, degrees are lowest which demonstrates subsets represent original best.