Insensitive Zone based Pinball Loss Twin Support Vector Machine for Pattern Classification

The Pinball error loss function has attracted wide research attention recently due to its excellent capability in handling feature noise present in data which eventually leads to robust classification models. Recently proposed Pinball loss based Support Vector Machine (Pin-SVM) [1] and Twin Support Vector Machine (Pin-TWSVM) [2] have shown effective experimental performances to ensure this claim. However, the main focus is to ensure robustness to noise and outliers, that has led to limited sparsity in the classifier(s) which result in the increased testing time and unstability in re-sampling. In this paper, we develop two robust Pin-TWSVM based models termed as εTWSVM and Flex-Pin-TWSVM that not only handles the noise present in data but also control model sparsity using userregulated and self-optimized insensitive zone respectively. The insensitive zone obtained in the model correspond to zero value of the dual variable and thus lead to simpler model. Extensive experiments performed on standard machine learning datasets confirm the efficacy of the proposed noise-insensitive models which also presents better sparseness than related traditional classifiers.