Most elementary statistical inference algorithms assume that the data can be modeled by a set of linear parameters with a normally distributed noise component. A new class of algorithms called support vector machine (SVM) remove such constraint. rea...