Experimental Prediction and Design of Ultra-Wideband Raman Amplifiers using Neural Networks

A machine learning method for Raman gain prediction and multi-pump broadband amplifier design is experimentally demonstrated over a 100 nm-wide optical bandwidth. We show high accuracy and ultra-fast prediction of arbitrary gain profile over a 100 km-long SSMF span.