Title :
A new approach of modifying SVM outputs
Author :
Madevska-Bogdanov, Ana ; Nikolic, Dragan
Author_Institution :
Fac. of Natural Sci. & Math., Skopje, Macedonia
Abstract :
We present a way of interpreting the outputs of support vector machine classifiers. Using the geometrical interpretation of the classifying hyperplane and the distance of the patterns from the hyperplane, we calculate the posterior probability, i.e. construct a rejection threshold for the vectors belonging to one of the classes
Keywords :
geometry; learning (artificial intelligence); neural nets; pattern classification; probability; classifying hyperplane; geometrical interpretation; posterior probability; rejection threshold; support vector machine classifiers; Geometry; Kernel; Mathematics; Polynomials; Probability; Support vector machine classification; Support vector machines; Training data;
Conference_Titel :
Neural Networks, 2000. IJCNN 2000, Proceedings of the IEEE-INNS-ENNS International Joint Conference on
Conference_Location :
Como
Print_ISBN :
0-7695-0619-4
DOI :
10.1109/IJCNN.2000.859427
