Backward Oversmoothing: why is it hard to train deep Graph Neural Networks?

Re-assign community