Using Simple Neural Networks to Correct Errors in Optical Data Transmission.

We have demonstrated the applicability of neural-network-based systems to the problem of reducing the effects of signal distortion, and shown that such a system has the potential to reduce the bit-error-rate in the digitized version of the analogue electrical signal derived from an optical data stream by a substantial margin over existing techniques.