Boosting in probabilistic neural networks

0410888 - UTIA-B 20020102 RIV US eng C - Conference Paper (international conference)
Grim, Jiří - Pudil, Pavel - Somol, Petr
Boosting in probabilistic neural networks.
Los Alamitos: IEEE Computer Society, 2002. ISBN 0-7695-1699-8. In: Proceedings of the 16th International Conference on Pattern Recognition. - (Kasturi, R.; Laurendeau, D.; Suen, C.), s. 136-139
[International Conference on Pattern Recognition /16./. Québec City (CA), 11.08.2002-15.08.2002]
R&D Projects: GA ČR GA402/01/0981; GA AV ČR KSK1019101
Institutional research plan: CEZ:AV0Z1075907
Keywords : neural networks * finite mixtures * boosting
Subject RIV: BB - Applied Statistics, Operational Research
http://library.utia.cas.cz/separaty/historie/grim-boosting in probabilistic neural networks.pdf

It has been verified in practical experiments that the classification performance can be improved by increasing the weights of misclassified training samples. We prove that in case of maximum-likelihood estimation the weighting of discrete data vectors is asymptotically equivalent to multiplication of the estimated distributions by a positive function. Consequently, the Bayesian decision-making can be made asymptotically invariant with respect to arbitrary weighting of data under certain conditions.
Permanent Link: http://hdl.handle.net/11104/0130975