A New Support Vector Data Description with Fuzzy Constraints

Abstract

This paper presents a novel approach to eliminate the effect of noisy samples from the learning step of Support Vector Data Description (SVDD) method. SVDD is a popular kernel method which tries to fit a hypersphere around the target object and can obtain more flexible and more accurate data descriptions by using proper kernel functions. Nonetheless, the SVDD could sometimes generate such a loose decision boundary while some noisy samples (outliers) exist in the training set. In order to solve this problem we define fuzzy constraints and two new concepts for each learning sample. Duo to the usage of fuzzy constraints, we called this method Fuzzy Constraints SVDD (FCSVDD). The overall experiments show prominence of our proposed method in comparison with the standard SVDD.