Publication details

Recognition of handwritten numerals by structural probabilistic neural networks

Conference Paper (international conference)

Grim Jiří, Pudil Pavel, Somol Petr


serial: Proceedings of the Second ICSC Symposium on Neural Computation, p. 528-534 , Eds: Bothe H., Rojas R.

publisher: ICSC, (Wetaskiwin 2000)

action: NC 2000, (Berlin, DE, 23.05.2000-26.05.2000)

research: AV0Z1075907

project(s): IAA2075703, GA AV, GA402/97/1242, GA ČR, VS96063, MŠMT

preview: Download

abstract (eng):

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.

Cosati: 12B, 09H, 06D

RIV: BB