Title: Learning Linear Sets Separation
Authors: Abdellatief Hussien AbouAli
Volume: 1
Issue: 9
Pages: 196-205
Publication Date: 2017/11/28//
Abstract:
Learning the separating hyper-planes between given sets of vectors have numerous application areas such as: segmentation, classifiers, and function approximation. In this paper, the weighted mean algorithm adapted to finding the separating hyper-plain between two sets is studied. The study was limited to two-set case and a single representative each. A proof of convergence proposed for the case of linear separable sets. Also, characterization to the algorithm in the low order nonlinear separable case behavior was part of the study. The experimental testing for linear separable sets proved that the algorithm convergence does not require large iterations. Also, in case of non-linear separable the algorithm has the tendency to be in minimal error states than others.