The Complexity of Explaining Neural Networks Through (Group) Invariants [pdf]

proceedings.mlr.press