Abstract:In this paper, we first analyzed the possible change of support vector set after new samples are added, then presented a new support vector machine incremental learning algorithm. This algorithm reconstructed SVM classifier through the selection of training samples in incremental learning based on change regularity of support vectors after new samples are added. Finally, the algorithm has a higher classification accuracy than traditional SVM incremental algorithms through experimental verification.

A New Incremental Learning Algorithm Based on Support Vector Machines