Electronic Neural Network Chips

Basic properties of matter will limit the miniaturization of transistors, the basic building block of computers, to dimensions only about ten times smaller than today's most advanced devices. Further advances in computing power will have to come from greater circuit integration and from new computing architectures. 

The research described in this lecture uses bio-physical models of brain function as points of departure for developing new classes of silicon integrated circuit chips. These chips, which combine analog and digital processing, excel in performing certain functions that are difficult to implement in conventional digital circuitry, particularly the task of associative recall.