If the receiver of an anonymous message is taken to be the adversary, then their best guess from a set of candidate senders ...

In this paper, we propose a versatile graph inference framework for learning from graph signals corrupted by exponential family noise. Our framework generalizes previous methods from continuous smooth ...