Fig. 3: The average MSE evaluated for the training and testing examples versus the iteration number for the various datasets.
From: Characterization and control of open quantum systems beyond quantum noise spectroscopy
The first row is for a CPMG_G_X_28 and b CPMG_S_X_28. The second row is for c CPMG_G_XY_7 and d CPMG_G_XY_pi_7. The final row is for e CPMG_G_XY_7_nl and f CPMG_G_XY_pi_7_nl. The plots show a “good fit” model since both lines are close for each datasets, and they are decreasing as the number of iterations increase. The small fluctuations at the end of the plots result from the gradients of the loss function being noisy, and are exaggerated due to the logarithmic scale of the plot.
