ONSET OF SUPER RETRIEVAL PHASE AND ENHANCEMENT OF THE STORAGE CAPACITY IN NEURAL NETWORKS OF NONMONOTONIC NEURONS

Analogue neural networks of associative memory with continuous time dy namics are studied for hon-monotonic transfer functions using the meth od of self-consistent signal-to-noise analysis. The Hebb learning rule with unbiased random patterns is assumed for the synaptic couplings. A novel phenomenon is found to occur as a result of a phase transition concerning the property of the local field distribution. In retrieval states of the newly found phase which we refer to as the super retrie val phase, noise in the local fields vanishes and the memory retrieval without errors ensues even for an extensive number of memory patterns stored under the local learning rule. The storage capacity is obtaine d as a function of the parameter representing the degree of non-monoto nicity of the transfer functions, with the result that an enhancement of the storage capacity can also occur.