0410298 20240103182204.420060210235959.9 On-line nonlinear estimation Praha ÚTIA AV ČR 1998 8 s. Preprint of the 3rd European IEEE Workshop on Computer-Intensive Methods in Control and Data Processing11-18 cav_un_auth*0101146 Kulhavý Rudolf UTIA-B Ústav teorie informace a automatizace AV ČR, v. v. i. 09I AV0Z1075907 The Bayesian identification of complex models is known to require extensive computer resources. Practical implementation requires approximation of the theoretically optimal solution. The paper discusses three major approaches to approximate Bayesian estimation-local weighting of data, reduction of model family to representative points and minimum relative entropy restoration of the information divergence of the empirical and model distributions of data. cav_un_auth*0212632 CMP '98 /3./ Praha CZ 25.07.1998-29.07.1998 BC AS http://hdl.handle.net/11104/0130389 UTIA-B 20000014 1998 1998 Praha ÚTIA AV ČR Preprint of the 3rd European IEEE Workshop on Computer-Intensive Methods in Control and Data Processing 11 18