Integration of Structural Expert Knowledge about Classes for Classification Using the Fuzzy Supervised Neural Gas

Marika Kästner, Wieland Hermann, Thomas Villmann
2012 The European Symposium on Artificial Neural Networks  
In this paper we describe an approach to integrate structural expert knowledge about class relations into classication schemes under the assumption of unary class coding. Exemplary, we show in a medical application such an integration for incorporation of prior medical knowledge about uncertainty for distinguishing patient classes. This knowlegde is integrated here in a class dissimilarity measure used for training the classication model.
dblp:conf/esann/KastnerHV12 fatcat:bof4ci7bxnbuhkxqims2ywmi2y