A neural network model using inter-pattern association (IPA)

Summary form only given, as follows. An investigation was conducted of a neural network model, the IPA (inter-pattern association) model, in which a basic logical operation is used to determine the inter-pattern association (i.e. association between the reference patterns), and a simple rule is applied to construct the tristate interconnections in the network. Computer simulations of the reconstruction of similar English letters in random noise have shown a better performance by the IPA neural network over the Hopfield model. A 2-D hybrid optical neural network is used to show the usefulness of this novel model. Since there are only three gray levels in the interconnection weight matrix, the requirement for the dynamic range of the spatial light modulators is easily satisfied, and the interconnections in the proposed network are much simpler than those in other models.<>