A modular YLSI design of a CMOS Hamming network

The authors describe a modular VLSI approach for the design and implementation of an optimum classifier Hamming network. A practical combination of digital and analog circuits is proposed to implement the neural network. The network consists of a digital single-layer perceptron followed by an analog winner-take-all circuit. The size and number of exemplar patterns that can be stored in the network are ideally unlimited due to the modularity of the design.<
>