ABSTRACT
In this chapter we present several synthesis methods that are used in the realization of associative memories by means of artificial recur rent neural networks. In the present section we first give an overview of the Outer Product Method and the Projection Learning Rule, while in Section 8.2 we present some extensions to the latter. In the fol lowing two sections, we develop a third synthesis method to realize associative memories, called the Eigenstructure Method, using linear systems operating on a closed hypercube in Section 8.3, and a variety of other recurrent neural network models in Section 8.4. In the next two sections, we devise yet another synthesis method for realizing associative memories, which is based on the Perceptron Training Al gorithm. A variety of recurrent neural network models are employed in Sections 8.5 and 8.6 in implementing this algorithm. Finally, in Section 8.7, a number of specific examples are considered to demon strate the usefulness of the results presented.
