Open AccessStochastic approximation algorithms for support vector machines semi-supervised binary classification
Author(s) -
Vaida Bartkutė-Norkūnienė
Publication year - 2008
Publication title -
lietuvos matematikos rinkinys
Language(s) - English
Resource type - Journals
eISSN - 2335-898X
pISSN - 0132-2818
DOI - 10.15388/lmr.2008.18114
Subject(s) - simultaneous perturbation stochastic approximation , support vector machine , binary classification , lipschitz continuity , differentiable function , perturbation (astronomy) , algorithm , operator (biology) , binary number , mathematical optimization , stochastic optimization , mathematics , stochastic approximation , convex function , computer science , regular polygon , artificial intelligence , stochastic process , computer security , repressor , mathematical analysis , chemistry , arithmetic , key (lock) , biochemistry , geometry , quantum mechanics , transcription factor , statistics , physics , gene
In this paper,we consider the problem of semi-supervisedbinary classificationby Support Vector Machines (SVM). This problem is explored as an unconstrained and non-smooth optimization task when part of the available data is unlabelled. We apply non-smooth optimization techniques to classification problems where the objective function considered is non-convex and non-differentiable and so difficult to minimize. We explore and compare the properties of Stochastic Approximation algorithms (Simultaneous Perturbation Stochastic Approximation (SPSA) with the Lipschitz Perturbation Operator, SPSA with the Uniform Perturbation Operator, and Standard Finite Difference Approximation) for semi-supervised SVM classification. We present some numerical results obtained by running the proposed methods on several standard test problems drawn from the binary classification literature.