EFFECT OF SYNAPSE DILUTION ON THE MEMORY RETRIEVAL IN STRUCTURED ATTRACTOR NEURAL NETWORKS

We investigate a simple model of structured attractor neural network ( ANN). In this network a module codes for the category of the stored in formation, while another group of neurons codes for the remaining info rmation. The probability distribution of stabilities of the patterns a nd the prototypes of the categories are calculated, for two different synaptic structures. The stability of the prototypes is shown to incre ase when the fraction of neurons coding for the category goes down. Th en the effect of synapse destruction on the retrieval is studied in tw o opposite situations : first analytically in sparsely connected netwo rks, then numerically in completely connected ones. In both cases the behaviour of the structured network and that of the usual homogeneous networks are compared. When lesions increase, two transitions are show n to appear in the behaviour of the structured network when one of the patterns is presented to the network. After the first transition the network recognizes the category of the pattern but not the individual pattern. After the second transition the network recognizes nothing. T hese effects are similar to syndromes caused by lesions in the central visual system, namely prosopagnosia and agnosia. In both types of net works (structured or homogeneous) the stability of the prototype is gr eater than the stability of individual patterns, however the first tra nsition, for completely connected networks, occurs only when the netwo rk is structured.