Support vector regression in the primal space based on the Ramp loss function

• 1、Heilongjiang Bayi Agricultural University,College of Sciences, Daqing 163319
• 2、Heilongjiang Bayi Agricultural University,College of Economics&Management ,Daqing 163319
• 3、Heilongjiang Bayi Agricultural University ,College of Sciences,Dqing 163319

Abstract：In this paper the function of support vector regression machine model relies much on the loss function chosen. On the basis of quadratic insensitive control type loss function Ramp with asymmetric form, support vector regression machine model is proposed. Adopting the concave and convex process optimization and the smooth technology algorithm, transforming the non-convex optimization to the continuous, quadratic differentiable convex optimization. Using the Amijo-Newton optimized algorithm to solve the proposed model and analyzing the convergence of the algorithm. The algorithm not only keeps the sparse nature of support vector, but also can control the abnormal values of the training sample. The experimental results show that the support vector regression machine model proposed keeps good generalization ability, and the model can better fit both the simulated data and the standard data. Compared with the standard support vector machine (SVM) model, the proposed model not only can reduce the effects of noise and outliers, but also has stronger robustness.

Keywords： theory of controls Support vector regression Outliers Loss function concave-convex procedure（CCCP）