Kazan (Volga region) Federal University, KFU
KAZAN
FEDERAL UNIVERSITY
 
A LINEAR SEPARABILITY CRITERION FOR SETS OF EUCLIDEAN SPACE
Form of presentationArticles in international journals and collections
Year of publication2013
Языканглийский
  • Gabidullina Zulfiya Ravilevna, author
  • Bibliographic description in the original language Gabidullina Z.R., A Linear Separability Criterion for Sets of Euclidean Space//Journal of Optimization Theory and Applications. - 2013. - Vol.158, Is.1. - P.145-171. (WOS)
    Annotation We prove new theorems which describe a necessary and sufficient condition for linear (strong and non-strong) separability and inseparability of the sets in a finite-dimensional Euclidean space. We propose a universal measure for the thickness of geometric margin (both the strong separation margin (separator) and the margin of unseparated points (pseudo-separator)) formed between parallel generalized supporting hyperplanes of two sets which are separated. The introduced measure permits one to compare results of linear separation obtained by different techniques for both disjoint and linearly inseparable sets. The optimization program the formulation of which provides for the separable sets a maximum thickness of the separator is considered. When the sets are inseparable, the same solver is guaranteed to construct a pseudo-separator with a minimum thickness. We estimate the distance between the convex closed sets. We construct a cone of generalized support vectors for hyperplanes each
    Keywords cone of support vectors, distance between the sets, separator, pseudo-separator, thickness of the separator (pseudo-separator), generalized supporting hyperplane, generalized support vector, projection
    The name of the journal Journal of Optimization Theory and Applications
    URL https://www.scopus.com/inward/record.uri?eid=2-s2.0-84878795454&partnerID=40&md5=042317a492d6982d1ec3ea8212359a11
    Please use this ID to quote from or refer to the card https://repository.kpfu.ru/eng/?p_id=148851&p_lang=2

    Full metadata record