Cancer sub-type detection using Modified-TSVM

A modification to Transductive SVM is implemented by selecting the filtered transductive samples. Filtering is done by considering the fact that the samples that fall with in the margin have richer information to find a better separating hyperplane. This algorithm outperformed the traditional SVM with an improvement on an average of 9% in accuracy when implemented on different cancer detection datasets.