Collective Computing Circuits Based on Neural Network Models

Biological computers, our brains, solve problems such as recognizing speech with little apparent effort, while electronic computers do this still very poorly. Presumably the brain does data processing massively in parallel and with a kind of analog computation. There has been a rapidly growing interest recently in models of neural structures to study whether they may lead to new, efficient algorithms and architectures to solve some of the problems biological systems do so well. Speech recognition, image analysis or reasoning based on a large volume of knowledge are examples of such tasks.