Meta-Learning Nonlinear Equalizer for Hyperparameters Tuning for 206.9Tbps PCS-WDM Optical Transmission

A meta-learning-based nonlinear equalizer architecture is proposed for hyperparameters tuning and experimentally demonstrated in a S+C+L-band PCS128QAM coherent optical transmission over 150km-G654.E, achieving an AIR of 206.9Tbps while significantly reducing additional training cost by 80.2%.