On generating kernels by fuzzy relations

Autoren Bernhard Moser
Titel On generating kernels by fuzzy relations
Typ Technischer Bericht
Nummer SCCH-TR-0527
Ort Hagenberg, Austria
Institution SCCH
Abteilung KVS
Jahr 2005
SCCH ID# 527
Abstract

This paper provides examples of fuzzy similarity relations which are proven to be positive-semidefinite and, therefore, kernels in the sense of Machine Learning. By this approach it is demonstrated how the semantics of fuzzy logic can be used for the construction of kernels.