A Attacks overview

Any improvements that are not entangled with a learning objective, i.e., pure pre-processing, usually arising from clues found in the node features and graph structure.