Abstract: We consider an improved model of the empirical risk minimization problem and its continuous relaxation. The continuous relaxation of the formulated problem is compared with the mathematical model used in the support vectors method. The results of numerical experiments comparing different models for problems with linearly inseparable sets are presented.

Keywords: cluster, decision rule, discriminant function, linear and nonlinear programming, nonsmooth optimization

ACM Classification Keywords: G.1.6 Optimization - Gradient methods, I.5 Pattern Recognition; I.5.2 Design Methodology - Classifier design and evaluation

Link:

A COMPARISON OF SOME APPROACHES TO THE RECOGNITION PROBLEMS IN CASE OF TWO CLASSES

Yurii I. Zhuravlev, Yuryi Laptin, Alexander Vinogradov, Aleksey Likhovid

http://www.foibg.com/ijima/vol02/ijima02-02-p01.pdf