An approach to optimal classification

Abstract

Classification of many objects is a problem which occurs both in industry and in the military. In complex industrial processes such as making plastic materials in a chemical factory, manufacturing electronics, building sensors for robots which are to be installed on the assembly line, or running nuclear plants, one is faced with an occasional failure. In order to correct it one must be able to identify a faulty component (or several) among the large number of components, by observing, often indirectly, the way the various components function. Similarly in military situations one is required to identify specific objects (e.g. ships or airplanes) among a large population of moving or still objects.