Universum Learning for SVM Regression

Sauptik Dhar, Vladimir Cherkassky

This paper extends the idea of Universum learning [18, 19] to regression problems. We propose new Universum-SVM formulation for regression problems that incorporates a priori knowl…