Mathematics

New Machine Learning Study Findings Reported from Singapore National University

  2012 APR 10 - (VerticalNews.com) -- "We investigate Newton-type optimization methods for solving piece-wise linear systems (PLSs) with nondegenerate coefficient matrix. Such systems arise, for example, from the numerical solution of linear complementarity problem, which is useful to model several learning and optimization problems," scientists writing in the journal Neural Computation report.

  "In this letter, we propose an effective damped Newton method, PLS-DN, to find the exact (up to machine precision) solution of nondegenerate PLSs. PLS-DN exhibits provable semiiterative property, that is, the algorithm converges globally to the exact solution in a finite number of iterations. The rate of convergence is shown to be at least linear before termination. We emphasize the applications of our method in modeling, from a novel perspective of PLSs, some statistical learning problems such as box-constrained least squares, elitist Lasso (Kowalski & Torreesani, 2008), and support vector machines (Cortes & Vapnik, 1995)," wrote X.T. Yuan and colleagues, Singapore National University.

  The researchers concluded: "Numerical results on synthetic and benchmark data sets are presented to demonstrate the effectiveness and efficiency of PLS-DN on these problems."

  Yuan and colleagues published their study in Neural Computation (Nondegenerate Piecewise Linear Systems: A Finite Newton Algorithm and Applications in Machine Learning. Neural Computation, 2012;24(4):1047-1084).

  Additional information can be obtained by contacting X.T. Yuan, Singapore National University, Dept. of Elect & Comp Engn, Singapore 117583, Singapore.

  The publisher of the journal Neural Computation can be contacted at: Mit Press, 55 Hayward Street, Cambridge, MA 02142, USA.

  Keywords: City:Singapore, Country:Singapore, Region:Asia, Cyborgs, Emerging Technologies

  This article was prepared by VerticalNews Mathematics editors from staff and other reports. Copyright 2012, VerticalNews Mathematics via VerticalNews.com.