Statistical, Logic-Based, and Neural Networks Based Methods for Mining Rules from Data

0404121 - UIVT-O 20020011 RIV NL eng C - Konferenční příspěvek (zahraniční konf.)
Holeňa, Martin
Statistical, Logic-Based, and Neural Networks Based Methods for Mining Rules from Data.
Multisensor Fusion. Dordrecht: Kluwer Academic Publishers, 2002 - (Hyder, A.; Shahbazian, E.; Waltz, E.), s. 511-532. NATO Science Series. ISBN 1-4020-0722-1.
[Conference of the NATO-Advanced-Study-Institute on Multisensor Data Fusion. Pitlochry (GB), 25.06.2000-07.07.2000]
Grant CEP: GA AV ČR IAB2030007
Výzkumný záměr: AV0Z1030915
Klíčová slova: data mining * integrative framework * observational logic * statistical hypotheses testing * rule extraction with artificial neural networks
Kód oboru RIV: BA - Obecná matematika

In the paper, a survey of the most important data mining approaches is given, and an integrative framework based on observational logic is described. Basic notions pertaining to the calculus of observational logic are explained, and it is shown how the calculus allows to capture data mining based on statistical hypotheses testing and on two approaches to the extraction of rules from data with artificial neural networks.
Trvalý link: http://hdl.handle.net/11104/0124391