Table 3 Average RMSE values for training dataset sizes of 0.1, 1, and 10 million samples. Significant improvement is observed when the sample size increases from 0.1 to 1 million, whereas only marginal improvement occurs with a further increase to 10 million.
From: Supervised learning of the Jaynes–Cummings Hamiltonian
Method | \(\epsilon\) | 0.1M | 1M | 10M | |||
|---|---|---|---|---|---|---|---|
Avg. | Energy | Avg. | Energy | Avg. | Energy | ||
vDNN | 0 | 4.18e-04 | 5.34e-03 | 5.01e-05 | 6.13e-04 | 1.07e-04 | 1.25e-03 |
0.4 | 1.45e-02 | 2.28e-02 | 1.31e-02 | 3.00e-02 | 1.43e-02 | 1.55e-02 | |
0.8 | 2.89e-02 | 5.34e-02 | 2.71e-02 | 6.75e-02 | 2.86e-02 | 3.61e-02 | |
dDNN | 0 | 2.43e-03 | 6.58e-03 | 1.58e-03 | 2.35e-03 | 1.48e-03 | 2.15e-03 |
0.4 | 4.11e-03 | 8.74e-03 | 3.04e-03 | 5.58e-03 | 2.89e-03 | 5.16e-03 | |
0.8 | 9.33e-03 | 1.42e-02 | 7.62e-03 | 1.15e-02 | 7.25e-03 | 1.05e-02 | |
