Improved information measurement and filter-based feature selection algorithm based on GA

摘要

To provide a technique for designing a filter type feature selection algorithm (IMGA) based on improved information measurement and GA, constructing a fitness function of a genetic algorithm by information measurement, Improve it so that it can be applied to selection, retrieve the optimal gene subset, and evaluate the classification effect of the selected feature subset using SVM as classifier. SOLUTION: The provided method is an effective feature selection method, and better classification performance can be realized as compared with the conventional method. BACKGROUND OF THE INVENTION