Abstract: Graph invariant learning (GIL) seeks invariant relations between graphs and labels under distribution shifts. Recent works try to extract an invariant subgraph to improve out-of-distribution ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results