Understanding deep neural networks is a major research objective with notable
experimental and theoretical attention in recent years. The practical success
of excessively large networks underscores the need for better theoretical
analyses and justifications. In this paper we focus on l