Dec. 20 2005 2:40PM

Dec. 20 2005 3:00PM

Efficient multiple-participant decision making relies on a cooperation of participants. Basic element of such cooperation is a mutual exchange of individual knowledge pieces, which are, in case of Bayesian decision makers, epresented by probability density functions. As individual participants may use completely different parameterized models, the knowledge can be exchanged only by sharing probability density functions of quantities describing data which are in common of communicating participants. In the presentation, a promising, technically feasible method, which allows a Bayesian decision maker to utilize an information in form of a probabilistic model of data will be proposed. The presented method is consistent with Bayesian learning from data - learning from an external probabilistic model is equivalent to the ordinary Bayesian learning, if the external model is described by an empirical distribution.

International Workshop on Data - Algorithms - Decision Making (Contact person: Martin Janžura)