"classification confidence; classifier aggregation; classifier combining; ensemble methods"@en . . "Static vs. Dynamic Classifier Systems in Classifier Aggregation" . "Spojov\u00E1n\u00ED klasifik\u00E1tor\u016F je metoda pro zlep\u0161en\u00ED kvality klasifikace - m\u00EDsto pou\u017E\u00EDv\u00E1n\u00ED jednoho klasifik\u00E1toru je vytvo\u0159en t\u00FDm klasifik\u00E1tor\u016F a v\u00FDstupy jednotliv\u00FDch klasifik\u00E1tor\u016F jsou pot\u00E9 agregov\u00E1ny pro z\u00EDsk\u00E1n\u00ED fin\u00E1ln\u00ED predikce. V\u011Bt\u0161ina metod pro agregaci klasifik\u00E1tor\u016F je statick\u00E1, tj. agregace se nep\u0159izp\u016Fsobuje konkr\u00E9tn\u00EDm klasifikovan\u00FDm vzor\u016Fm. V tomto \u010Dl\u00E1nku pop\u00ED\u0161eme dynamick\u00E9 syst\u00E9my klasifik\u00E1tor\u016F, kter\u00E9 pou\u017E\u00EDvaj\u00ED koncept dynamick\u00E9 konfidence klasifikace, aby se p\u0159izp\u016Fsobily konkr\u00E9tn\u00EDmu vzoru. V\u00FDsledky experiment\u016F na 4 um\u011Bl\u00FDch a 4 re\u00E1ln\u00FDch datov\u00FDch mno\u017Ein\u00E1ch ukazuj\u00ED, \u017Ee dynamick\u00E9 syst\u00E9my mohou dosahovat signifikantn\u011B lep\u0161\u00EDch v\u00FDsledk\u016F ne\u017E statick\u00E9 syst\u00E9my."@cs . "Porovn\u00E1n\u00ED statick\u00E9 a dynamick\u00E9 agregace klasifik\u00E1tor\u016F"@cs . "397265" . . "Classifier aggregation is a method for improving quality of classification -- instead of using just one classifier, a team of classifiers is created, and the outputs of the individual classifiers are aggregated into the final prediction. Common methods for classifier aggregation are static, i.e., they do not adapt to the currently classified pattern. In this paper, we introduce a formalism of dynamic classifier systems, which use the concept of dynamic classification confidence to dynamically adapt to the currently classified pattern. Results of experiments with quadratic discriminant classifiers on four artificial and four real-world benchmark datasets show that dynamic classifier systems can significantly outperform static classifier systems."@en . "RIV/68407700:21340/08:04150956!RIV09-MSM-21340___" . . . . . "1"^^ . "Static vs. Dynamic Classifier Systems in Classifier Aggregation"@en . "RIV/68407700:21340/08:04150956" . "21340" . . "Z(MSM6840770039)" . "978-80-01-04195-6" . "Static vs. Dynamic Classifier Systems in Classifier Aggregation"@en . . "Static vs. Dynamic Classifier Systems in Classifier Aggregation" . "[368BF4B1C6E7]" . "1"^^ . . . "2008-11-07+01:00"^^ . "Praha" . "Praha" . "11"^^ . "Porovn\u00E1n\u00ED statick\u00E9 a dynamick\u00E9 agregace klasifik\u00E1tor\u016F"@cs . "Classifier aggregation is a method for improving quality of classification -- instead of using just one classifier, a team of classifiers is created, and the outputs of the individual classifiers are aggregated into the final prediction. Common methods for classifier aggregation are static, i.e., they do not adapt to the currently classified pattern. In this paper, we introduce a formalism of dynamic classifier systems, which use the concept of dynamic classification confidence to dynamically adapt to the currently classified pattern. Results of experiments with quadratic discriminant classifiers on four artificial and four real-world benchmark datasets show that dynamic classifier systems can significantly outperform static classifier systems." . . "\u0160tefka, David" . . . "\u010Cesk\u00E1 technika - nakladatelstv\u00ED \u010CVUT" . . . . "Doktorandsk\u00E9 dny 2008" . .