large margin cellular piecewice linear classifier

Piecewise linear classifiers have attracted a lot ofattention in recent years, because of their simplicity andclassification capability. in this paper, a large margin cellularpiecewise linear classifier is introduced, called cell-svm. thecellular structure of cell-svm obtains a piecewise linear decisionboundary which handles non-linearly separable data. unlike theconventional svm approaches, the proposed method employsmulti hyperplanes instead of one in search space and resultingcellular structure addresses some important issues in machinelearning such as: multi-modal classes, nonlinear classification,noisy data and outliers, small sample size, multi-class classificationand overfitting to training samples. in experiments, wedemonstrate significant gains for the well-known benchmark realdatasets when compared to the usual multi-class svm techniqueswith rbf kernel like ovo svm, ova svm and mc-svm.besides, it is shown that the proposed method achieves comparableresults to other popular classification methods such as neuralnetwork and decision tree which performs better in general.