CNN : A Brainlike Supercomputer on a Chip
The CNN (Cellular Nonlinear Networks) is a programmable computer on a silicon chip capable of executing 10 billion signal-processing operations per second. This immense computing power is achieved via a non-Von Neumann architecture where information is processed as flows towards a sequence of attractors where nonlinear dynamics are coded to mimic brainlike morphological functions. This lecture presents an introduction to the principles and applications of a novel paradigm shift in computing based on a locally-connected array of simple locally-active cells where emergence from attractors provides the information being sought. Applications will be drawn from various areas, including the tracking of missiles, detection of plasma instabilities, video information processing, neuromorphic function simulation, etc.