Feb. 5 2008 2:00PM

Feb. 5 2008 3:00PM

Bayesian approach is a powerful tool for data processing. It is often applied to the filtration, prediction and control tasks. All mentioned tasks require estimation of the system parameters. The tool is originally intended for processing of consistent data arising from systems with constant parameters. Estimation of time varying parameters is much more difficult. This case is mostly solved by using estimation of constant parameters combined with a kind of forgetting. The main principle of the forgetting is that the older data have the smaller weight in estimation than the newer ones. This lecture will give an insight into the problem of optimal choice of the forgetting rate using advanced forgetting technique respecting partially time-varying parameters.

Pravidelný seminář Rozhodování a řízení za neurčitosti. (Kontaktní osoba: Miroslav Kárný)