Measure-Theoretic Anti-Causal Representation Learning

Arman Behnam, Binghui Wang

Causal representation learning in the anti-causal setting (labels cause features rather than the reverse) presents unique challenges requiring specialized approaches. We propose An…