Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

The piecewise linear classifier DIPOL92

Der stückweise lineare Klassifikator DIPOL92
: Schulmeister, B.; Wysotzki, F.

Bergando, F.; Raedt, L. de:
ECML '94. European Conference on Machine Learning. Proceedings
Berlin: Springer, 1994 (Lecture Notes in Artificial Intelligence)
ISBN: 3-540-57868-4
ISSN: 0302-9743
European Conference on Machine Learning <1994, Catania>
Conference Paper
Fraunhofer IITB, Außenstelle Prozessoptimierung (EPO); 1997
machine learning; maschinelles Lernen; piecewise linear classification; stückweise lineare Klassifikation

This paper presents a learning algorithm which constructs an optimised piecewise linear classifier for nclass problems. In the first step of the algorithm initial positions of the discriminating hyperplanes are determined by linear regression for each pair of classes. To optimise these positions depending on the misclassified patterns an error criterion function is defined. This function is minimised by a gradient descent procedure for each hyperplane separately. As an option in the case of non-convex classes, a clustering procedure decomposing the classes into appropriate subclasses can be applied. The classification of patterns is defined on a symbolic level on the basis of the signs of the discriminating hyperplanes.