Modular neural associative memory capable of storage of large amounts of data

A new neural net architecture based on the Hopfield network is proposed. This architecture overcomes the memory limitation that is peculiar to a single network at the cost of moderate computational expenses. Parameters' influence on read-write processes is considered, possible read errors are defined and estimations for associative recall effectiveness as a function of search complexity are given. Theoretical estimations are in close correspondence with experimental results obtained for random vectors dataset.