Abstraction hierarchy in deep learning neural networks.

Roman Ilin, Thomas Watson, Robert Kozma

Browse the full IJCNN paper archive.