Classifier optimization method using niche genetic algorithm
基于小生境遗传算法的分类优化方法

According to multi-classify problem, the multi-classes classifier constructed by binary classes classifier are usually very slow to be trained. When a large number of categories of data are to be classified, the training work could be very difficult. Hyper-sphere support vector machine (HSSVM) can be extended to solve this multi-classification problem. Each category data trains only one HSSVM, the sample space is divided by multiple optimal hyper-spheres. In order to improve the performance of classifier, this paper used improved crowding niche genetic algorithm (ICNGA) to select features, chose optimal feature subset for different target classes. Using UCI data set of numerical experiment shows that the classifiers have a higher accuracy if ICNGA has been used for feature selection, especially the sample data has a large number of categories or feature vectors.