A Classifier Research Based on RS Reducts and SVM Ensemble
Guojun Zhang · 2010
In order to construct a high-performance ensemble classifier, it needs that the basic classifiers, which contained by the ensemble one, have higher classification precision and their classification error is independent from each other. In fact, it is too difficult to choose these basic classifiers satisfying the two conditions above. Rough reduction is the core in the fields of Rough Set theory. Each reduct not only contains lesser attributes, but also is independent from each other in classification error. SVM is a promising method of machine learning based on the structural risk minimization principle, which has high classification precision. In the paper, an ensemble classifier algorithm based on RS reducts and SVM is provided in order to construct a high-performance classifier. Experiment shows that our proposed algorithm is efficient and has better classification accuracy and better performance.