A dynamically reconfigurable M-SIMD implementation architecture for large scale neural-digital hybrid processing

A modular, reconfigurable, parallel and linearly scalable hardware implementation architecture for realization of large scale neural networks, called the modular neural ring (MNR), has been developed, prototyped, and shown to be highly effective in hardware implementation of large scale neural computing models. The authors examine the possibility of extending the use of the architecture to vector digital processing by taking advantage of its parallelism and the modular reconfigurability. This hybrid neural-digital computing architecture has been tested and found to offer a uniform hardware platform for highly parallel, modular, and reconfigurable implementations of both digital and neural processing tasks