. "5"^^ . . "Optimalizace parametr\u016F u\u010Den\u00ED a architektury neuronov\u00FDch s\u00EDt\u00ED pomoc\u00ED evolu\u010Dn\u00EDch algoritm\u016F"@cs . . . "P\u0159\u00EDsp\u011Bvek se zab\u00FDv\u00E1 problematikou optimalizace parametr\u016F u\u010Den\u00ED a architektury neuronov\u00FDch s\u00EDt\u00ED pomoc\u00ED evolu\u010Dn\u00EDch algoritm\u016F. Optimalizovan\u00E1 s\u00ED\u0165 byla testov\u00E1na p\u0159i kr\u00E1tkodob\u00E9 p\u0159edpov\u011Bdi pr\u016Ftok\u016F (6 h) na povod\u00ED horn\u00EDho toku \u0159eky S\u00E1zavy. Pou\u017Eit\u00FDm typem neuronov\u00E9 s\u00EDt\u011B byl v\u00EDcevrstevn\u00FD perceptron s u\u010Den\u00EDm zp\u011Btnou propagac\u00ED chyby. Po optimalizaci parametr\u016F byly neuronov\u00E9 s\u00EDt\u011B natr\u00E9nov\u00E1ny a byly provedeny simulace. Kvalita p\u0159edpov\u011Bdi byla hodnocena vybran\u00FDmi krit\u00E9rii. Z v\u00FDsledk\u016F vypl\u00FDv\u00E1, \u017Ee optimalizovan\u00E9 neuronov\u00E9 s\u00EDt\u011B maj\u00ED p\u0159i kr\u00E1tkodob\u00E9 p\u0159edpov\u011Bdi dobr\u00E9 v\u00FDsledky. Optimalizace parametr\u016F p\u0159isp\u00EDv\u00E1 ke zlep\u0161en\u00ED kvality p\u0159edpov\u011Bdi a m\u016F\u017Ee b\u00FDt vyu\u017Eita pro p\u0159esn\u011Bj\u0161\u00ED volbu hodnot parametr\u016F ovliv\u0148uj\u00EDc\u00EDch u\u010Den\u00ED a simulace."@cs . . "Optimalizace parametr\u016F u\u010Den\u00ED a architektury neuronov\u00FDch s\u00EDt\u00ED pomoc\u00ED evolu\u010Dn\u00EDch algoritm\u016F"@cs . "1"^^ . . "1"^^ . . . . . "neural networks, evolutionary algorithms, runoff forecast, optimization"@en . . "277269" . . "Vodohospod\u00E1\u0159sk\u00E9 technicko-ekonomick\u00E9 informace" . "Havl\u00ED\u010Dek, Vojt\u011Bch" . . "I" . . . "RIV/60460709:41330/10:47929!RIV11-MZE-41330___" . "52" . "[933442273606]" . "Optimalizace parametr\u016F u\u010Den\u00ED a architektury neuronov\u00FDch s\u00EDt\u00ED pomoc\u00ED evolu\u010Dn\u00EDch algoritm\u016F" . "The paper deals with artificial neural networks learning and architecture parameters optimization by evolutionary algorithms. Optimized network was tested by short-term runoff forecasts (6 h) on the S\u00E1zava upper river basin. Type of used neural network was the multilayer perceptron with back propagation learning algorithm. The learning and the simulations were performed after optimizing the parameters of the neural network. The efficiency of the predictions was evaluated by selected criteria. The results show that optimized neural networks provide good results for short-term forecast. Optimization of parameters helps to improve the efficiency of forecasts and may be used for more precise choice of parameter values that affect learning and simulation."@en . "RIV/60460709:41330/10:47929" . "P\u0159\u00EDsp\u011Bvek se zab\u00FDv\u00E1 problematikou optimalizace parametr\u016F u\u010Den\u00ED a architektury neuronov\u00FDch s\u00EDt\u00ED pomoc\u00ED evolu\u010Dn\u00EDch algoritm\u016F. Optimalizovan\u00E1 s\u00ED\u0165 byla testov\u00E1na p\u0159i kr\u00E1tkodob\u00E9 p\u0159edpov\u011Bdi pr\u016Ftok\u016F (6 h) na povod\u00ED horn\u00EDho toku \u0159eky S\u00E1zavy. Pou\u017Eit\u00FDm typem neuronov\u00E9 s\u00EDt\u011B byl v\u00EDcevrstevn\u00FD perceptron s u\u010Den\u00EDm zp\u011Btnou propagac\u00ED chyby. Po optimalizaci parametr\u016F byly neuronov\u00E9 s\u00EDt\u011B natr\u00E9nov\u00E1ny a byly provedeny simulace. Kvalita p\u0159edpov\u011Bdi byla hodnocena vybran\u00FDmi krit\u00E9rii. Z v\u00FDsledk\u016F vypl\u00FDv\u00E1, \u017Ee optimalizovan\u00E9 neuronov\u00E9 s\u00EDt\u011B maj\u00ED p\u0159i kr\u00E1tkodob\u00E9 p\u0159edpov\u011Bdi dobr\u00E9 v\u00FDsledky. Optimalizace parametr\u016F p\u0159isp\u00EDv\u00E1 ke zlep\u0161en\u00ED kvality p\u0159edpov\u011Bdi a m\u016F\u017Ee b\u00FDt vyu\u017Eita pro p\u0159esn\u011Bj\u0161\u00ED volbu hodnot parametr\u016F ovliv\u0148uj\u00EDc\u00EDch u\u010Den\u00ED a simulace." . "Artificial Neural Networks Learning and Architecture Parameters Optimization by Evolutionary Algorithms"@en . "Artificial Neural Networks Learning and Architecture Parameters Optimization by Evolutionary Algorithms"@en . "P(QH91247)" . "0" . "0322-8916" . "CZ - \u010Cesk\u00E1 republika" . "41330" . . "Optimalizace parametr\u016F u\u010Den\u00ED a architektury neuronov\u00FDch s\u00EDt\u00ED pomoc\u00ED evolu\u010Dn\u00EDch algoritm\u016F" .