modelLoss.m

function [lossG,lossD,gradientsG,gradientsD,stateG,scoreG,scoreD] = ...
    modelLoss(netG,netD,X,T,Z,flipFactor)

% Calculate the predictions for real data with the discriminator network.
YReal = forward(netD,X,T);

% Calculate the predictions for generated data with the discriminator network.
[XGenerated,stateG] = forward(netG,Z,T);
YGenerated = forward(netD,XGenerated,T);

% Calculate probabilities.
probGenerated = sigmoid(YGenerated);
probReal = sigmoid(YReal);

% Calculate the generator and discriminator scores.
scoreG = mean(probGenerated);
scoreD = (mean(probReal) + mean(1-probGenerated)) / 2;

% Flip labels.
numObservations = size(YReal,4);
idx = randperm(numObservations,floor(flipFactor * numObservations));
probReal(:,:,:,idx) = 1 - probReal(:,:,:,idx);

% Calculate the GAN loss.
[lossG, lossD] = ganLoss(probReal,probGenerated);

% For each network, calculate the gradients with respect to the loss.
gradientsG = dlgradient(lossG,netG.Learnables,RetainData=true);
gradientsD = dlgradient(lossD,netD.Learnables);

end