AI Paper Atlas

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

A Simple Feature Augmentation for Domain Generalization

Pan Li, Da Li, Wei Li, Shaogang Gong, Yanwei Fu, Timothy M. Hospedales

ICCV 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Foundation-model domain generalization · 220 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 multiplicative noise to additive noise (SFA-S beta -> beta+alpha) (ablation)SFA-S · Digit-DGavg accuracy: 77.0 -> 79.2Improved
add adaptive noise xi to alpha and beta (ablation)SFA-A · Digit-DGavg accuracy: 79.2 -> 79.6Improved
use multiplicative noise only vs additive only (ablation)SFA-S · Digit-DGavg accuracy: 77.0 -> 78.3Improved
use Normal instead of Laplace noise (ablation)SFA-S(beta) · Digit-DGavg accuracy: 76.8 -> 77.0Improved
use Normal instead of uniform noise (ablation)SFA-S(beta) · Digit-DGavg accuracy: 76.2 -> 77.0Improved
inject noise at Layer 2 instead of Layer 0 (ablation)SFA-S · Digit-DGavg accuracy: 76.8 -> 79.2Improved
vary noise strength sigma from 1 to 1000 (ablation)SFA-S · Digit-DGavg accuracy: 79.2 vs 78.2 at sigma 1000No clear effect
add SFA feature augmentation to Vanilla ERMVanilla ERM · Digit-DGavg accuracy: 73.7 -> 79.6Improved
add SFA-A to EISNetEISNet · PACSavg accuracy: 81.9 -> 82.5Improved
add SFA-A to DDAIGDDAIG · PACSavg accuracy: 81.0 -> 81.8Improved
use adaptive SFA-A instead of SFA-SSFA · PACSavg accuracy: 81.0 -> 81.7Improved
use adaptive SFA-A instead of SFA-SSFA · VLCSavg accuracy: 73.5 -> 74.0Improved

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