An incremental convex hull algorithm based online Support Vector Regression

Abstract

Consider the time complexity and newly added samples, an incremental convex hull algorithm based online Support Vector Regression (ICH-OSVR) is proposed in this paper, which can significantly reduce the time consuming and realize fast online learning when added a new sample. There are two steps, called offline step and online step. Firstly, the convex hull vertices of training samples are selected by using convex hull offline algorithm and then regard the vertices of convex hull as the training samples, which are prepared for training. Secondly, when a new sample comes and it is out of the previous convex hull, update the vertices of convex hull and then the previous SVR model will be updated by the new convex hull, but if the new sample is within the previous convex hull, discard it and do not need to update the model. The effectiveness of our proposed methods has been confirmed according to the artificial data sets and real data sets.