Discriminant analysis based on exponential possibility distributions

The paper deals with an exponential possibility distribution and its application to discriminant analysis. The proposed discriminant analysis is formulated by minimizing the possibility or the necessity measure when two possibility distributions are given. This formulation can be reduced to the well-known eigenvalue problem. An unknown input can be classified by the proposed discriminant rule. Furthermore, this discriminant analysis is extended to the case where a set of several unknown inputs is given