Sufficient Invariant Learning for Distribution Shift

Taero Kim, Subeen Park, Sungjun Lim +3

Learning robust models under distribution shifts between training and test datasets is a fundamental challenge in machine learning. While learning invariant features across environ…