Reproduce “Partial Invariance for Classification under Structural Causal Models” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Austin Goddard, Yu Xiang
The authors introduce dIMP, a classifier for predicting labels in a new environment using labeled data from other environments and unlabeled test features. It estimates class probabilities by combining invariant expectations for target-child features with quantities estimated from test features. In the light-tunnel experiment, its neural-network implementation outperforms considered baselines in all three test-environment choices, under invariance and invertibility assumptions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.