Support Vector Machine (SVM) currently aroused great attention from machine learning community. It offers a new learning strategy to obtain optimal hyperlane and yields a high generalization to the unseen pattern. It also opens a possibility to incorporate prior knowledge of the domain by designing a specific kernel function to measure the similarity degree of the data. These features explained why in several studies, SVM generalized better than the confentional classification methods including the neural network trained by backpropagation algorithm. Keywordsi: Support Vector Machine, Soft Computing, Neural Network, Margin, Biomedical Data