Abstract: Graph neural networks are designed to learn functions on graphs. Typically, the relevant target functions are invariant with respect to actions by permutations. Therefore the design of some ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results