Least squares support vector machine regression with boundary condition

Regression plays an important role in signal processing, identifying and modeling. This paper proposes a regression algorithm based on least squares support vector machine. In the algorithm, the equality constraints without errors term are adopted at the point with boundary condition. The equality constraints without errors term force the regression model to pass through the given special points and satisfy boundary condition. The algorithm is applied to sine function regression and good performances are obtained. The proposed algorithm provides a new attempt for regression with boundary condition.