Improvements on twin-hypersphere support vector machine using local density information

Abstract

In this paper, we propose a novel binary classifier called twin-hypersphere support vector machine with local density information (LDTHSVM). Firstly we extract local density for each training sample and treat it as the weight of that sample, next prune training dataset according to these local density degrees, finally introduce these local density degrees into twin-hypersphere support vector machine (THSVM) and reconstruct classification model on the pruned training dataset. LDTHSVM not only inherits good properties from THSVM, but also gives more robust description for dataset. The experimental results on synthetic and publicly available benchmark datasets show the excellent performance of the LDTHSVM classifier in terms of classification accuracy and learning time.