Details:
Summary:
Several applications simulating human intelligence, which demand unconventional computing have been explored. The applications will serve to introduce some approaches to the study of new computing architectures. In reviewing them, one can try to distinguish among the terms parallel, analog, and neural. Computational properties of the neuronal synapse have been outlined. A simple, hybrid, electronic model of neuron-like computing element has been presented. Important features of its performance have been compared to data from biological neurons involved in visual information processing. Then, the question of how large numbers of the computing elements might be organized have been revisited. In conclusion, a consideration is made of some of the hurdles which must be passed to bring the new technologies out of the laboratory and into cost-effective commercial applications.