Massive memory organizations for implementing neural networks

摘要

A single-input multiple-data architecture which has n processing elements and n/sup 2/ memory modules arranged in an n*n array is presented. This massive memory is used to store the weights of the neural network being simulated. It is shown how networks with sparse connectivity among neurons can be simulated in O( square root n+e) time. where n is the number of neurons and e the number of interconnections in the network. Preprocessing is carried out on the connection matrix of the sparse network resulting in data movement that has an optimal asymptotic time complexity and a small constant factor.