A Learning Machine in the Context of the General Control Problem

Advances in automatic control theory may be seen to have two main objectives: the synthesis of improved controllers for a given plant and the extension of the range of plants considered for control. Optimal control theory and the application to linear systems is an example of the former, whilst advances in the latter have been largely concerned with describing-function and quasi-linearization techniques for the linear approximation of non-linear but reasonably continuous plant. It has become conventional to introduce the abstract concept of a plant through its state-transitions