Machine learning-based EDFA Gain Model Generalizable to Multiple Physical Devices

We report a neural-network based erbium-doped fiber amplifier (EDFA) gain model built from experimental measurements. The model shows low gain-prediction error for both the same device used for training ($MSE \leq 0.04\ dB^{2}$) and different physical units of the same make (generalization $MSE\leq 0.06\ dB^{2}$).