Incremental learning and forgetting in RBF networks and SVMs with applications to financial problems

Abstract

Radial Basis Function Networks (RBFNs) have been widely applied to practical classification problems. In recent years, Support Vector Machines (SVMs) are gaining much popularity as promising methods for classification problems. This paper compares those two methods in view of incremental learning and forgetting. The authors have reported that the incremental learning and active forgetting in RBFNs provide a good performance for classification under the changeable environment. First in this paper, a method for incremental learning and forgetting in SVMs is proposed. Next, a comparative simulation for a portfolio problems between RBFNs and SVMs will be made.