AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Invariant and Transportable Representations for Anti-Causal Domain Shifts

Yibo Jiang, Victor Veitch

NeurIPS 2022 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Optimization-based domain generalization · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
ACTIR instead of ERMACTIR · Synthetictest accuracy: 9.95 -> 74.77Improved
fine-tune ACTIR on 10 labeled examplesACTIR · Syntheticaccuracy: 74.77 -> 89.28Improved
fine-tune IRM on 10 labeled examplesIRM · Syntheticaccuracy: 74.91 -> 74.27No clear effect
ACTIR instead of IRMACTIR · Color MNISTtest accuracy: 59.97 -> 70.30Improved
fine-tune ACTIR on 10 labeled examplesACTIR · Color MNISTaccuracy: 70.30 -> 85.25Improved
ACTIR instead of IRMACTIR · Camelyon17test accuracy: 71.59 -> 77.73Improved
ACTIR instead of IRMACTIR · PACS Artaccuracy: 78.69 -> 82.55Improved
ACTIR instead of IRMACTIR · PACS Sketchaccuracy: 65.61 -> 62.14Hurt
ACTIR instead of ERMACTIR · VLCS Caltechaccuracy: 95.29 -> 95.15No clear effect
ACTIR instead of MAMLACTIR · VLCS SUN09accuracy: 70.84 -> 72.31Improved
ACTIR instead of MAMLACTIR · VLCS Caltechaccuracy: 96.75 -> 95.15Hurt
data violates anti-causal structure: ACTIR instead of IRMACTIR · Synthetic (non anti-causal)test accuracy: 69.61 -> 43.51Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy