AI Paper Atlas

Research map / Learning theory and algorithms / Kernels and regression theory

Model-Based Domain Generalization

Alexander Robey, George J. Pappas, Hamed Hassani

NeurIPS 2021 · Learning theory and algorithms › Kernels and regression theory · approach: Empirical generalization measures · 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
enforce prediction invariance to learned domain transformations with primal-dual trainingMBDG · Camelyon17-WILDS OOD validationaccuracy: 73.3 ± 9.9 -> 94.8 ± 0.4Improved
enforce prediction invariance to learned domain transformations with primal-dual trainingMBDG · FMoW-WILDS OOD validationaccuracy: 51.3 (0.4) -> 52.3 ± 0.5Improved
enforce invariance to learned image transformationsMBDG · ColoredMNIST Avgaccuracy: 40.7 -> 68.5Improved
enforce invariance to learned image transformationsMBDG · ColoredMNIST -90%accuracy: 23.1 ± 10.7 -> 63.5 ± 0.0Improved
enforce prediction invariance using learned domain transformationsMBDG · PACS Avgaccuracy: 83.7 -> 85.6Improved
enforce prediction invariance using learned domain transformationsMBDG · PACS Sketch (S)accuracy: 78.6 ± 1.4 -> 85.2 ± 0.2Improved
replace primal-dual invariance training with ERM plus generated-image augmentation (ablation)MBDG · ColoredMNIST Avgaccuracy: 68.5 -> 48.3Hurt
use generated-image augmentation in the training objective instead of primal-dual constraint enforcement (ablation)MBDG · ColoredMNIST Avgaccuracy: 68.5 -> 59.0Hurt
replace adaptive dual updates with a fixed regularization weight (ablation)MBDG · ColoredMNIST Avgaccuracy: 68.5 -> 58.1Hurt
use a fixed regularization weight rather than dual ascent (ablation)MBDG · ColoredMNIST G-invariance constraint satisfactiondistRegHurt

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