Can I, as a user, get insight in the exact functions MATLAB uses for accuracy and RMSE calculation in the trainNetwork function for volumetric outputs?
I try to compare differnt training progress plots. 2 networks are trained with MATLAB's trainNetwork function, several other CNNs are custom training loops. All work on volumetric images. Input and out but have a dlarray format of 'SSSCB'. The segmentation labels pixels in 3 categories.