AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

RealPatch: A Statistical Matching Framework for Model Patching with Real Samples

Sara Romiti, Christopher Inskip, Viktoriia Sharmanska, Novi Quadrianto

ECCV 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Group-robust classification · 0 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 matched real-sample consistency regularisation (RealPatch vs SGDRO)SGDRO · CelebArobust accuracy (%): 82.96 -> 84.82Improved
use RealPatch instead of SGDROSGDRO · CelebArobust gap (%): 7.13 -> 5.19Improved
use RealPatch instead of CAMELCAMEL · CelebArobust gap (%): 5.09 -> 5.19No clear effect
use real matched pairs instead of CycleGAN counterfactuals (RealPatch vs CAMEL)CAMEL · Waterbirdsrobust accuracy (%): 76.82 -> 84.44Improved
use RealPatch instead of GDROGDRO · Waterbirdsaggregate accuracy (%): 88.26 -> 86.89Hurt
use RealPatch instead of GDROGDRO · Waterbirdsrobust gap (%): 14.80 -> 4.43Improved
use RealPatch instead of SGDROSGDRO · Waterbirdsrobust accuracy (%): 83.11 -> 84.44Improved
use RealPatch instead of ERMERM · iWildCam-smallrobust gap (%): 19.65 -> 11.36Improved
use RealPatch instead of SGDROSGDRO · iWildCam-smallrobust accuracy (%): 75.50 -> 76.70Improved
add temperature scaling (t<1) to propensity scores (ablation)RealPatch Stage 1 matching · iWildCam-small covariate balance (SMD<=0.1 count)covariates with SMD <= 0.1: 753 -> 1125Improved
add fixed caliper to matching (ablation)RealPatch Stage 1 matching · iWildCam-small covariate balance (SMD<=0.1 count)covariates with SMD <= 0.1: 1037 -> 1125Improved
match with RealPatch instead of original imSitu dataResNet activity classifier · imSitu (211 classes)model leakage: 76.79 -> 68.76Improved

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