A Maximal Margin Hypersphere SVM

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


In this work we propose a generalization of the Support Vector Machine (SVM) method in which the separator is a curve, but the concept of margin and maximization of the margin is still present. The idea of using different functions for the separation has been explored in particular in the scope of hyperspheres. However, most of these proposals use two spheres, using concepts different from maximal margin or one sphere but with a poor performance when data from the classes have a linear shape. In this paper we present a formulation of the linear SVM that generalizes it to a spherical separation shape, but still maximizing the margin. A linear relaxation of this quadratic formulation is also presented. The performance of these two formulations for classification purpose is tested and the results are encouraging.

Original languageEnglish
Title of host publicationComputational Science and Its Applications – ICCSA 2021: 21st International Conference, Cagliari, Italy, September 13–16, 2021, Proceedings, Part V
EditorsOsvaldo Gervasi, Beniamino Murgante, Sanjay Misra, Chiara Garau, Ivan Blečić, David Taniar, Bernady O. Apduhan, Ana Maria Rocha, Eufemia Tarantino, Carmelo Maria Torre
Place of PublicationCham
Number of pages16
ISBN (Electronic)978-3-030-86976-2
ISBN (Print)978-3-030-86975-5
Publication statusPublished - 2021
Event21st International Conference on Computational Science and Its Applications, ICCSA 2021 - Virtual, Online
Duration: 13 Sept 202116 Sept 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12953 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference21st International Conference on Computational Science and Its Applications, ICCSA 2021
CityVirtual, Online


  • Automatic classification
  • Non-linear SVM
  • SVM


Dive into the research topics of 'A Maximal Margin Hypersphere SVM'. Together they form a unique fingerprint.

Cite this