Research On Decision Theory Based On The Multi-Prior Expected Utility

Decision analysis mostly researches on uncertainty, it's conventional decision model is subject expected utility theory (SEU). Being perspicuity, subject expected utility model has been an important theory to analyze how to make decision under uncertainty. But the real world is Complicated, people can't construct a unique prior to describe uncertainty exactly, So SEU model is faulty in real decision-making. Aiming at the phenomenon, some people have improved on it's axioms systems, such as choquet expected utility model of Schmeidler , multi-prior expected utility model of Gilboa and Schmeidler. Multi-prior expected utility model is one of better uncertainty decision theorys and now widely applied at home and abroad, which dosen't require a unique subject probability distribution.Based on the Bayesian decision theory, Expected utility representation with multiple prior is discussed in this paper, and new criteria are constructed, which can extend the use range of the old criteria and deal with uncertainty decision problems better. Detailed contents are followed:Firstly, basic decision theories are presented, such as utility theory, decision principle. The weakness of SEU model are pointed out, we find a way of making use of multi-prior model to solve the decision problems better.Secondly, expected utility representation with multiple prior is elaborated;new criteria are constructed, we find that the new criteria can be used widely than the old one.Lastly, the new criteria are applied to a decision problem.