AI Paper Atlas

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

Overcoming Shortcut Learning in a Target Domain by Generalizing Basic Visual Factors from a Source Domain

Piyapat Saranrittichai, Chaithanya Kumar Mummadi, Claudia Blaiotta, Mauricio Munoz, Volker Fischer

ECCV 2022 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Statistical alignment and hypothesis testing · 7 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
add source domain with IL constraintFactor-0 · DiagVib-AnimalHM accuracy: 1.5 -> 41.3Improved
add source domain with IL constraintFactor-0 · Color-FruitHM accuracy: 5.5 -> 57.0Improved
add source domain with IL constraintFactor-0 · AO-CLEVRHM accuracy: 4.3 -> 32.1Improved
add source domain with IL constraintFactor-0 · Color-FashionHM accuracy: 3.6 -> 22.3Improved
use factor instead of global representation with source (ablation)FactorSRC · Color-FruitHM accuracy: 2.2 -> 42.4Improved
add isolated latent constraint to FactorSRC (ablation)FactorSRC · DiagVib-AnimalHM accuracy: 13.0 -> 41.3Improved
add cross-factor independence constraint to FactorSRC (ablation)FactorSRC · Color-FruitHM accuracy: 42.4 -> 19.6Hurt
add IL constraint (ablation)FactorSRC · DiagVib-Animal cross-prediction z_a to objectcross-prediction accuracy (lower is better): 77 -> 46Improved
add regularization when learning association matrix (ablation)FactorSRC-IL-LA · DiagVib-Animal (semi-correlated)HM accuracy: 7.1 -> 34.7Improved
make source factors uncorrelated (ablation)FactorSRC-IL · DiagVib-AnimalHM accuracy: 2.5 -> 33.5Improved
use uncorrelated Caltech source instead of correlated target data (ablation)FactorSRC-IL · DiagVib-AnimalHM accuracy: 1.7 -> 33.5Improved
use Caltech shapes instead of MNIST shapes in source (ablation)FactorSRC-IL · Color-FruitHM accuracy: 46.9 -> 56.0Improved

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