Fig. 5: PINN convergence histories and cross-validations against acoustic measurements from excised syrinx experiments.
From: Predicting 3D soft tissue dynamics from 2D imaging using physics informed neural networks
a The history of data (left) and equation (right) loss during PINN training for one of the syringeal models. Data loss and PINN loss converged after around 27,000 and 70,000 epochs, respectively. b Comparison of acoustic parameters between PINN prediction and experimental data. The bar shows the mean value, and the error bar represents the standard deviation for four subjects. The gray straight lines indicate the correspondence of each subject between experimental measurement and PINN prediction. Individual data points are marked as red circles.
