Sparse \(L_0\)-norm least squares support vector machine with feature selection (Q6544594)

The paper is devoted to the ``least squares support vector machine'' model proposed to find a hyperplane that divides a given set of vectors into two classes. To improve the efficiency on small size data sets it is developed a new ``sparse least squares support vector machine'' model reducing the number of parameters in the corresponding optimization function. The authors adopt ``alternating direction method of multiples'' to solve the new model. They prove the convergence of the designed algorithm and provide its complexity analysis. The numerical performance of the algorithm is demonstrated on simulated and benchmark data sets.