Discovering causally invariant features for out-of-distribution generalization

Yujie Wang, Kui Yu, Guodu Xiang, Fuyuan Cao, Jiye Liang

Abstract: Highlights•We propose a CIFD framework to find accurate causal variables for OOD generalization.•Potential causal variables are identified by a double-layer local causal structure.•True causal variables are learnt by a double-layer total causal effect estimator.•Comprehensive experiments demonstrate the superiority of CIFD over the SOTA methods.