A stochastic logic neural network as a deterministic and probabilistic Hopfield network

Summary form only given. The information processing of a stochastic logic neural network, which is one of the pulse-coded artificial neural network families, was investigated. This network realizes pseudo-analog performance with some local learning rules by using a digital circuit, and therefore it suits silicon technology. The limited synaptic weights reduce coding noise and suppress the degradation of memory storage capacity. To study the effect of coding noise on the optimization problem, the authors simulated a probabilistic Hopfield model, which has a continuous neuron output function and probabilistic behavior, with this architecture. The proper choice of unscheduled or scheduled coding noise improved the solutions of the traveling salesman problem. This result suggests that the stochastic logic may be useful for implementing probabilistic dynamics as well as deterministic dynamics