Incremental Learning Algorithm for Support Vector Data Description

Abstract

Support vector data description (SVDD) has become a very attractive kernel method due to its good results in many novelty detection problems.Training SVDD involves solving a constrained convex quadratic programming,which requires large memory and enormous amounts of training time for large-scale data set.In this paper,we analyze the possible changes of support vector set after new samples are added to training set according to the relationship between the Karush-Kuhn-Tucker (KKT) conditions of SVDD and the distribution of the training samples.Based on the analysis result,a novel algorithm for SVDD incremental learning is proposed.In this algorithm,the useless sample is discarded and useful information in training samples is accumulated.Experimental results indicate the effectiveness of the proposed algorithm.