Recognition of handwritten numerals by structural probabilistic neural networks

0410327 - UTIA-B 20000043 RIV CA eng C - Konferenční příspěvek (zahraniční konf.)
Grim, Jiří - Pudil, Pavel - Somol, Petr
Recognition of handwritten numerals by structural probabilistic neural networks.
Wetaskiwin: ICSC, 2000. ISBN 3-906454-22-3. In: Proceedings of the Second ICSC Symposium on Neural Computation. - (Bothe, H.; Rojas, R.), s. 528-534
[NC 2000. Berlin (DE), 23.05.2000-26.05.2000]
Grant CEP: GA ČR GA402/97/1242
Grant ostatní: GA AV(CZ) IAA2075703; MŠMT(CZ) VS96063
Program: IA
Výzkumný záměr: AV0Z1075907
Kód oboru RIV: BB - Aplikovaná statistika, operační výzkum
http://library.utia.cas.cz/separaty/historie/grim-recognition of handwritten numerals by structural probabilistic neural networks.pdf

The well known "beauty defect" of probabilistic neural networks is the biologically unnatural complete interconnection of neurons. Despite of deep reasons of this undesirable property, it can be removed by a special subspace approach without leaving the exact framework of Bayesian decision-making. The structural optimization based on EM algorithm is controlled by an infor- mation criterion. In the present paper the method has been applied to recognize unconstrained handwritten numerals.
Trvalý link: http://hdl.handle.net/11104/0130418