Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Comparison of a novel combined ECOC strategy with different multiclass algorithms together with parameter optimization methods

: Hülsmann, M.; Friedrich, C.M.

Postprint urn:nbn:de:0011-n-587947 (372 KByte PDF)
MD5 Fingerprint: 8929d343ad133604b65be90e2ba55c30
The original publication is available at
Erstellt am: 16.9.2010

Perner, P.:
Machine Learning and Data Mining in Pattern Recognition. 5th International Conference, MLDM 2007 : Leipzig, Germany, July 18-20, 2007, Proceedings
Berlin: Springer, 2007 (Lecture Notes in Artificial Intelligence 4571)
ISBN: 3-540-73498-8
ISBN: 978-3-540-73498-7
International Conference on Machine Learning and Data Mining in Pattern Recognition (MLDM) <5, 2007, Leipzig>
Konferenzbeitrag, Elektronische Publikation
Fraunhofer SCAI ()

In this paper we consider multiclass learning tasks based on Support Vector Machines (SVMs). In this regard, currently used methods are One-Against-All or One-Against-One, but there is much need for improvements in the field of multiclass learning. We developed a novel combination algorithm called Comb-ECOC, which is based on posterior class probabilities. It assigns, according to the Bayesian rule, the respective instance to the class with the highest posterior probability. A problem with the usage of a multiclass method is the proper choice of parameters.
Many users only take the default parameters of the respective learning algorithms (e.g. the regularization parameter C and the kernel parameter γ). We tested different parameter optimization methods on different learning algorithms and confirmed the better performance of One-Against-One versus One-Against-All, which can be explained by the maximum margin approach of SVMs.