Microarray data classification based on principal curves

In this paper, a novel classifier is proposed to classify microarray data using principal curves. Principal curves are the non-linear generalization of principal components. Intuitively, a principal curve `passes through the middle of the data cloud´. As a kind of new classification technique, Principal Curve-based classifier (PC) involves a novel way of computing a principal curve for each class using the training data. A test sample is given the class-label of the principal curve that is closest to it according to Expected Squared Error. Experimental results illustrate the performance of the PC is better than other existing approaches when a very small sample size is concerned.