The causality between data features has become a powerful tool for addressing out-of-distribution generalization issues, with its core idea being to identify invariant causal features across data ...