Institute of Information Theory and Automation

Sips of SIP

Lecturer: Ing. Miroslav Kárný, DrSc.
Date and time: 02.02.2015 - 11:00
Room: 474
Department: Adaptive Systems (AS)


The project SIP "Fully Probabilistic Design of Dynamic Decision Strategies for Imperfect Participants within Market Scenarios" has the following motivation and meta-aim: Decision making (DM) is a targeted selection among possible actions. Dynamic DM generates an actions' sequence using available knowledge and trying to respect preferences of the DM participant. Efficiency of a DM heavily depends on the adopted design of the action-generating strategies. Mostly, DM is made under uncertainty and incomplete knowledge. Then, under general conditions, the Bayesian DM maximising expected utility over admissible strategies should be used. Experience shows that real DM participants employ the prescriptive Bayesian DM theory much less than desirable from consistency and efficiency view points. Ubiquity of DM and its state determine importance of the considered research, which intends to offer to real imperfect DM participants a support that allows them to use a feasible extension of the prescriptive Bayesian DM theory even within the demanding market scenarios. The talk will outline state of the research oriented towards this meta aim with the stress on novel achievements and open problems seen.
Institute of Information Theory and Automation