AI Paper Atlas

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

On Calibration and Out-of-Domain Generalization

Yoav Wald, Amir Feder, Daniel Greenfeld, Uri Shalit

NeurIPS 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Optimization-based domain generalization · 9 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
Fine-tune top layers with CLOvEERM · FMoW OOD testworst unseen region accuracy: 32.63 -> 44.16Improved
Use robust isotonic calibrationERM · FMoW OOD testworst unseen region accuracy: 32.63 -> 37.19Improved
Fine-tune top layers with CLOvEDeepCORAL · Camelyon17 OOD testaccuracy: 72.44 -> 79.96Improved
Use robust isotonic calibrationDeepCORAL · Camelyon17 OOD testaccuracy: 72.44 -> 76.8Improved
Use naive isotonic calibrationERM · PovertyMap OOD testPearson correlation r: 0.832 -> 0.827Hurt
Use robust isotonic calibrationIRM · PovertyMap OOD testPearson correlation r: 0.735 -> 0.815Improved
Calibrate using CLOvEERM · CivilComments testaverage group accuracy: 63.65 -> 80.39Improved
Calibrate using CLOvEIRM · CivilComments testaverage group accuracy: 40.61 -> 68.45Mixed
Select models by average ECE across training environmentsColored MNIST models · Colored MNIST OOD evaluationPearson correlation with OOD accuracy: -0.92Improved
Use the IRMv1 penalty rather than average ECE for model selectionColored MNIST models · Colored MNIST OOD evaluationPearson correlation with OOD accuracy: -0.59Mixed
Optimize multi-domain calibration with CLOvE instead of IRMv1Two-Bit environments · Two-Bit OOD testinvariant classifier recoveryImproved

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