Fig. 7. Evolution of estimated unknown parameters and loss function in case 1 and case 5 during the training process.

Case 1 (A and B) involves geometry identification, while case 5 (C and D) involves both material and geometry identification. See Fig. 2 for the definitions of the cases. (A and C) The dashed lines and solid lines represent the reference value and estimated value of unknown parameters. Unknown parameters are not updated in the pretraining process during the first 20k (for case 1) and 50k (for case 5) iterations, respectively. (B and D) The value of the loss function during the training process. The results for cases 0, 2, 3, and 4 are shown in section S4 (fig. S1).