Introducing the ordered weighted average in the soft margin SVM

  1. Luisa I. Martínez-Merino
  2. Alfredo Marín
  3. Justo Puerto
  4. Antonio Manuel Rodriguez-Chia
Konferenzberichte:
31st European Conference on Operational Research (EURO 2021)

Verlag: EURO – the European Association of Operational Research Society

ISBN: 978-618-85079-1-3

Datum der Publikation: 2021

Seiten: 149

Art: Konferenz-Beitrag

GOOGLE SCHOLAR

Zusammenfassung

Support vector machines (SVMs) have become one of the most useful mathematical programming approaches for supervised classification.The classical soft-margin SVM model minimizes an objective function given by the inverse of the margin between the supporting hyperplanesand the sum of the deviations of misclassified objects penalized by a parameter. In this talk, we propose an SVM model where weights are assigned to the sorted values of slack variables associated with the deviations. Thus, we include the ordered weighted average operator in the softmargin SVM. Unlike other approaches, this is a one-step method where the classical model is adequately modified.