Minimax Decision-rule Concepts

Abstract

In real situations parameters such as a priori probabilities are rarely known exactly. In some problems a priori probabilities can be estimated with a high degree of confidence. In situations where this cannot be done the minimax criterion is often an acceptable alternative for determining the best receiver operation. The minimax decision function corresponds to the Bayes decision function for the a priori probabilities which makes the Bayes risk a maximum. The minimax criterion has the additional characteristic of being independent of the actual a priori probabilities.