Under weak overlap, a latent confounder can be approximately identified from a single proxy or multiple unlabeled sources, and a reweighted mixture-of-experts model adapts to confounder shift.
A causal framework for distribution generalization
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Scalable Out-of-distribution Robustness in the Presence of Unobserved Confounders
Under weak overlap, a latent confounder can be approximately identified from a single proxy or multiple unlabeled sources, and a reweighted mixture-of-experts model adapts to confounder shift.