This is the continuation of the previous blog post. Only difference is that we change the loss function as below:
\[ loss = MSEloss(z-v) + \lambda \times L1loss(z)\]
Summary:
| sigma of f | \(\lambda \) | learning rate | number of epochs | batch size | Q1 | Q2 | Q1 per bubble | Q2 per bubble | % gap between training and validation | |
| exp1 | 1 | 0.005 | 2e-5 | 300 | 1 | 13.431614 | 0.474453 | 0.531866 | 0.018680 | 58.3 |
| epx2 | 2 | 0.01 | 2e-5 | 300 | 1 | 20.737143 | 1.180204 | 0.821757 | 0.046736 | 4.9 |
Experiment 1: learning rate=2e-5, sigma of kernel f=1, regularizer parameter=0.005
Percentage gap between training and validation error ( \( \frac{validation errror – training error}{training error}*100 \) ) = 58.3
Average quantitative metric1 is 13.431614
Average metric1 per bubble is 0.531866
Average quantitative metric2 is 0.474453
Average metric2 per bubble is 0.018680
Worst Case:Center Location(x,z) = 29.6065 mm and 48.8026 mm , Average Metric1 per bubble: 0.874942, Average Metric2 per bubble: : 0.052237
Best Case:Center Location(x,z) =9.894499999999999 mm, 39.9322 mm, Average Metric1 per bubble: 0.483357, Average Metric2 per bubble: : 0.014553
Experiment 2: learning rate=2e-5, sigma of kernel f=2, regularizer parameter=0.01
Percentage gap between training and validation error ( \( \frac{validation errror – training error}{training error}*100 \) ) = 4.9
Average quantitative metric1 is 20.737143
Average metric1 per bubble is 0.821757
Average quantitative metric2 is 1.180204
Average metric2 per bubble is 0.046736
Worst Case:Center Location(x,z) = 29.6065 mm and 48.8026 mm , Average Metric1 per bubble: 0.955968, Average Metric2 per bubble: : 0.059352
Best Case:Center Location(x,z) =6.1985 mm, 36.4826 mm, Average Metric1 per bubble: 0.798583, Average Metric2 per bubble: : 0.043044
