. "Neruda, Roman" . . . . . "[B7BF67AB86C3]" . "P(GA526/03/Z042), S" . "Academia" . . "Application of artificial neural networks in modelling the hydrological balance of the Plou\u010Dnice River valley, Northern Bohemia (Czech Republic)" . . "1"^^ . "2"^^ . "rainfall-runoff models; Plou\u010Dnice River valley; artificial neural networks; runoff prediction"@en . "Application of artificial neural networks in modelling the hydrological balance of the Plou\u010Dnice River valley, Northern Bohemia (Czech Republic)"@en . "Doubice" . "RIV/44555601:13520/07:00003729!RIV08-MSM-13520___" . "13520" . . "Neruda, Martin" . . . "Application of artificial neural networks in modelling the hydrological balance of the Plou\u010Dnice River valley, Northern Bohemia (Czech Republic)" . . "Praha" . "Sandstone Landscapes" . "Prok\u00E1zali jsme, \u017Ee v\u00EDcevrstv\u00E9 perceptronov\u00E9 neuronov\u00E9 s\u00EDt\u011B mohou b\u00FDt \u00FAsp\u011B\u0161n\u011B pou\u017Eity pro vytvo\u0159en\u00ED mal\u00FDch sr\u00E1\u017Eko-odtokov\u00FDch model\u016F. Tyto modely mohou zpracov\u00E1vat pouze historick\u00E9 \u010Dasov\u00E9 \u0159ady. V\u0161eobecn\u011B jednovrstv\u00E9 s\u00EDt\u011B l\u00E9pe vystihuj\u00ED jednodenn\u00ED historii, zat\u00EDmco dvouvrstv\u00E9 s\u00EDt\u011B jsou \u00FAsp\u011B\u0161n\u011Bj\u0161\u00ED pro dvoudenn\u00ED historii. Vytvo\u0159en\u00E1 trojvrstv\u00E1 s\u00ED\u0165 l\u00E9pe p\u0159edpov\u00EDdala p\u0159i jednodenn\u00ED historii dat. Domn\u00EDv\u00E1me se, \u017Ee je to zp\u016Fsoben\u00E9 t\u00EDm, \u017Ee v\u011Bt\u0161\u00ED neuronov\u00E9 s\u00EDt\u011B funguj\u00ED l\u00E9pe pro v\u011Bt\u0161\u00ED datov\u00E9 \u0159ady. Na druhou stranu, s\u00EDt\u011B s v\u00EDce parametry maj\u00ED tendenci k p\u0159etr\u00E9nov\u00E1n\u00ED a tedy maj\u00ED men\u0161\u00ED chybu p\u0159i tr\u00E9ninku, ale v\u011Bt\u0161\u00ED chybu p\u0159i testov\u00E1n\u00ED."@cs . . . "RIV/44555601:13520/07:00003729" . . "Aplikace um\u011Bl\u00FDch neuronov\u00FDch s\u00EDt\u00ED na modelov\u00E1n\u00ED hydrologick\u00E9 bilance v povod\u00ED Plou\u010Dnice, v severn\u00EDch \u010Cech\u00E1ch (\u010Cesk\u00E1 republika)"@cs . "2002-01-01+01:00"^^ . . "Application of artificial neural networks in modelling the hydrological balance of the Plou\u010Dnice River valley, Northern Bohemia (Czech Republic)"@en . "978-80-200-1577-8" . "410422" . "Aplikace um\u011Bl\u00FDch neuronov\u00FDch s\u00EDt\u00ED na modelov\u00E1n\u00ED hydrologick\u00E9 bilance v povod\u00ED Plou\u010Dnice, v severn\u00EDch \u010Cech\u00E1ch (\u010Cesk\u00E1 republika)"@cs . . "90-92" . "3"^^ . . "We have shown that multilayer perceptron neural networks can be successfully used for creating small rainfall-runoff models. These models can be built from historical time series data only. In general, one layer networks fit the %22one day history%22 model better, while two layer networks are more succesful for the %22two days' history%22 nodels. The designed three layers network yielded better forecast on the %22one day history%22 data. We believe this is because bigger neural networks function better for large data sets. On the other hand, networks with more parameters tend to be over-trained and thus they achieve a smaller training error, but a larger testing error."@en . "We have shown that multilayer perceptron neural networks can be successfully used for creating small rainfall-runoff models. These models can be built from historical time series data only. In general, one layer networks fit the %22one day history%22 model better, while two layer networks are more succesful for the %22two days' history%22 nodels. The designed three layers network yielded better forecast on the %22one day history%22 data. We believe this is because bigger neural networks function better for large data sets. On the other hand, networks with more parameters tend to be over-trained and thus they achieve a smaller training error, but a larger testing error." . .