AN EFFICIENT CANCER CLASSIFICATION USING WEIGHTED KMEANS AND WEIGHTED SUPPORT VECTOR MACHINE

Abstract

Cancer classification is very important in the field of bioinformatics for diagnosis of cancer cells. Accurate prediction of cancer is very important for providing better treatment and to avoid the additional cost associated with wrong therapy. In recent years for classifying the cancer numbers of methods have been exist. The main objective is to find the smallest set of genes by using machine learning algorithms. The proposed method initially uses Support Vector Machine (SVM) classifier with great flexibility. However, the SVM is not suitable for the classification of large datasets because of significant computational problems. The SVM combined with the k-means clustering (Km-SVM) is a fast algorithm developed to accelerate both the training and the prediction of SVM classifiers by using the cluster centers obtained from the k-means clustering. The new techniques namely weighted K-means-SVM is implemented to improve accuracy and to reduce misclassification and noise arising from irrelevant genes. The proposed algorithm was compared with different classifier algorithms which were applied on the same database. The experimental results showed the superiority of the proposed algorithm that could achieve a classification accuracy of 85%