. "Application of neural networks in identification of nonlinear non-Gaussian systems is treated. Stress is laid on a parameter estimation of the networks. They are trained by the Gaussian sum method which is a global filtering method allowing to determine conditional probability density functions of network weights. Proposed approach to estimation of network weights (parameters) based on Gaussian sum filtering method overcomes commonly used prediction error methods and it is an interesting alternative to sequential Monte Carlo methods. The considered training approach is demonstrated by an illustration example." . "system identification; nonlinear non-Gaussian stochastic system; non-Gaussian disturbance; neural network training; Gaussian sum; Bayesian relations; multilayer perceptron network"@en . "Identifikace neline\u00E1rn\u00EDch negaussovsk\u00FDch syst\u00E9m\u016F neuronov\u00FDmi s\u00EDt\u011Bmi"@cs . "Oxford" . . . "Identifikace neline\u00E1rn\u00EDch negaussovsk\u00FDch syst\u00E9m\u016F neuronov\u00FDmi s\u00EDt\u011Bmi"@cs . "\u0160imandl, Miroslav" . "RIV/49777513:23520/05:00000357!RIV07-MSM-23520___" . "524128" . "3"^^ . "Elsevier" . "Identification of nonlinear non-gaussian systems by neural networks" . "Nonlinear control systems 2004" . "\u010Cl\u00E1nek je zam\u011B\u0159en na probl\u00E9m aplikace neuronov\u00FDch s\u00EDt\u00ED v identifikaci neline\u00E1rn\u00EDch negaussovsk\u00FDch syst\u00E9m\u016F. D\u016Fraz je kladen na odhad parametr\u016F neuronov\u00E9 s\u00EDt\u011B, kter\u00E9 jsou tr\u00E9nov\u00E1ny vyu\u017Eit\u00EDm metody Gaussovsk\u00FDch sm\u011Bs\u00ED, co\u017E je jedna z glob\u00E1ln\u00EDch filtra\u010Dn\u00EDch metod, kter\u00E9 umo\u017E\u0148uj\u00ED ur\u010Dit pravd\u011Bpodobnostn\u00ED hustotn\u00ED funkci vah s\u00EDt\u011B. Navr\u017Een\u00FD postup odhadu parametr\u016F (vah) s\u00EDt\u011B zalo\u017Ee\u00FD na metod\u011B Gaussovsk\u00FDch sm\u011Bs\u00ED p\u0159ekon\u00E1v\u00E1 obvykle pou\u017E\u00EDvan\u00E9 metody chyby predikce a p\u0159edstavuje zaj\u00EDmavou alternativu k sekven\u010Dn\u00EDm metod\u00E1m Monte Carlo. Navr\u017Een\u00FD p\u0159\u00EDstup tr\u00E9nov\u00E1n\u00ED je demonstrov\u00E1n v ilustra\u010Dn\u00EDm p\u0159\u00EDklad\u011B."@cs . "6"^^ . . . "3"^^ . "Application of neural networks in identification of nonlinear non-Gaussian systems is treated. Stress is laid on a parameter estimation of the networks. They are trained by the Gaussian sum method which is a global filtering method allowing to determine conditional probability density functions of network weights. Proposed approach to estimation of network weights (parameters) based on Gaussian sum filtering method overcomes commonly used prediction error methods and it is an interesting alternative to sequential Monte Carlo methods. The considered training approach is demonstrated by an illustration example."@en . "Identification of nonlinear non-gaussian systems by neural networks" . . . . "Identification of nonlinear non-gaussian systems by neural networks"@en . "Kr\u00E1l, Ladislav" . "RIV/49777513:23520/05:00000357" . "Identification of nonlinear non-gaussian systems by neural networks"@en . . . "[40D22BECA897]" . . "Hering, Pavel" . . . . . . . "1307-1312" . . . . "0-08-044303-6" . . "23520" . "Z(MSM 235200004)" .