Title :
Performance characterization of K-winner machines
Author :
Ridella, Sandro ; Zunino, Rodolfo
Author_Institution :
Dept. of Biophys. & Electron. Eng., Genoa Univ., Italy
Abstract :
The paper reports on new findings about the properties of K-winner machines (KWMs). The resulting theoretical model is sharply characterized in terms of generalization performance, and exhibits interesting features from an application perspective as well. The major novel aspect lies in connecting analytically the KWM framework to established methods, proposed by Vapnik and Cherkassky, for assessing a classifier´s generalization performance
Keywords :
generalisation (artificial intelligence); learning (artificial intelligence); neural nets; pattern classification; vector quantisation; K-winner machines; VC dimension; Vapnik expression; generalization; learning; pattern classification; performance; vector quantisation; Algorithm design and analysis; Calibration; Design optimization; Error analysis; Joining processes; Optical wavelength conversion; Performance analysis; Prototypes; Testing; Yield estimation;
Conference_Titel :
Neural Networks, 2001. Proceedings. IJCNN '01. International Joint Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-7044-9
DOI :
10.1109/IJCNN.2001.939536
