Neural networks and genetic algorithms as programming paradigm for a new CMOS-array computer

Artificial neural networks (ANN) and genetic algorithms (GA) have turned out to play an important role in building and programming of parallel computers. Especially fine granulated massively parallel systems, hard to control with conventional parallel programming languages, can be used easiliy for solving optimization problems and processing of high data volume, when a genetic optimizer or an ANN emulator shield the user from the computer sophisticated hardware. To demonstrate this we designed a flexible hardware capable of emulating ANNs. This includes different types of neurons including complex biologically motivated models based on activity pulses, variable pulse transmission times, and multiple threshold learning rules, as well as simple Sigma-Pi-units useful in image processing applications for filtering and image transformation.