We had learned in the lectures that a Neural Style Transfer network learns the generated image pixel values. But how are the NN parameters learned (i.e. W, b) ?
VGG19 is trained on imagenet