Accelerating TDECQ Assessments using Convolutional Neural Networks

We experimentally demonstrate the use of convolutional neural networks to accelerate TDECQ assessments for 400G direct-detect transmitter qualification. The method estimates TDECQ from static eye-diagrams ~1000 times faster than conventional methods with <; 0.25dB mean discrepancy.