We propose a new methodology for learning the structure of sparse networks from data; in
doing so we adopt a dual perspective where we consider networks both as weighted graphs …