On Large Deviation Principles for Large Neural Networks - Joan Bruna