NLSSVM: Least Square Support Vector Machine based on Newton optimization

摘要

The traditional optimization problem of Least Square Support Vector Machines (LSSVM) is solved by the linear equations that are time-consuming. In order to reduce the time-consuming, a novel algorithm called NLSSVM (LSSVM based on Newton optimization) is proposed in this paper. Firstly, NLSSVM converted the optimization problem of LSSVM to unconstrained optimization problem, then solved by Newton iterative optimized method. The experimental results on several real datasets indicate that NLSSVM can reduce the training time greatly without degrading the generalization ability of LSSVM, as compared with the traditional LSSVM.