Adaptive logic for artificially intelligent systems

The state-of-art of adaptive logic systems consisting of memory elements is surveyed with particular reference to three recent developments. The first introduces the concept of special-purpose structures of mixtures of memory elements and fixed circuitry for the implementation of several logic functions. The second discusses techniques developed for the automatic classification of patterns by means of randomly connected combinational networks and the optimization of performance through reconnexion of the network after training. The third topic describes the use of sequential systems for the computation of global properties of binary patterns.