Classification of Microarray Gene Expression Data Using a New Binary Support Vector System

We have developed a new system to classify microarray data. The system, which we call a binary support vector system (BSVS), is based on the use of support vectors in support vector machines (SVM) and binary classification to analyze microarray data. In this paper, the accuracy of BSVS is evaluated and compared with two well-known and established SVM systems: mySVM and LIBSVM. Our results show BSVS to be as accurate as mySVM and LIBSVM. BSVS might be preferable to the other two systems when analyzing gene expression, because it is simple in concept, has low computational complexity, is nearly free of parameter and kernel selection, and allows for a greater variety of definitions of similarity.