Feature Selection with Hybrid Mutual Information and Genetic Algorithm
Feature selection plays an important role in data mining and pattern recognition, especially for large scale data. During past years, various metrics have been proposed to measure the relevance between different features. Mutual information is nonlinear and can effectively represent the dependencies of features. In this paper, the authors proposed a combinatorial algorithm that uses the powerful metric mutual information and genetic algorithm. In this method, relevant features search with genetic algorithm by mutual information as fitness function. Totally, this method has better results than original mutual information in more situations and some of them the result of hybrid mutual information and genetic algorithm is equal with original mutual information.