An Evolutionary Gene Selection Method for Microarray Data Based on SVM Error Bound Theories

Microarrays have thousands to tens of thousands of gene features but patient samples are fewer or a few hundred. Identifying genes whose disruption causes congenital or acquired disease is the fundamental problem in microarray data analysis. In this paper, we propose an efficient evolutionary SVM-based classifier that can select smaller number of features with high accuracy. The proposed method uses SVM with a given subset of features to evaluate the fitness function, and new subset of features are selected based on several leave-one-out error bounds for the SVM classifier and the frequency of occurrence of the features in the evolutionary approach. We test our proposed method on different microarray data and find that the proposed method can obtain high classification accuracy with a smaller number of selected genes.