We show that adversarial training is unnecessary and sometimes counter-productive; we instead cast invariant representation learning asasingle information-theoretic objectivethat can bedirectly optimized.

This hasbeen shown to be insufficient for generating optimal representation for classification, and to find conditionally invariant representations, usually strong assumptions are needed.

Moreover,compellingevidenceshowsthat GNNs are extremely vulnerable to graph data shifts [79,26,20].

Machine learning models will often fail when deployed in an environment with a data distribution that is different than the training distribution.