AI Paper Atlas

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

BiaSwap: Removing Dataset Bias With Bias-Tailored Swapping Augmentation

Eungyeup Kim, Jihyeon Lee, Jaegul Choo

ICCV 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Data augmentation and generalization bias · 58 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 separation of bias-contrary and bias-guiding pairs (c1) (ablation)BiaSwap · Colored MNIST unbiased (99.0%)accuracy: 43.04 -> 86.03Improved
add separation of bias-contrary and bias-guiding pairs (c1) (ablation)BiaSwap · bFFHQ unbiasedaccuracy: 45.0 -> 58.87Improved
add CAM-based patch sampling (c2) (ablation)BiaSwap · Colored MNIST unbiased (99.0%)accuracy: 84.16 -> 86.03Improved
add CAM-based patch sampling (c2) (ablation)BiaSwap · bFFHQ unbiasedaccuracy: 51.2 -> 58.87Improved
replace BiaSwap with Stylised ImageNet augmentationvanilla classifier · Colored MNIST unbiased (99.0%)accuracy: 48.76 -> 40.79Hurt
use BiaSwap augmentation instead of SINclassifier · bFFHQ unbiased (99.0%)accuracy: 69.86 -> 78.98Improved
use BiaSwap augmentation instead of vanilla trainingclassifier · Colored MNIST unbiased (99.0%)accuracy: 48.76 -> 83.74Improved
use BiaSwap instead of LfFLfF · Corrupted CIFAR10 unbiased (99.5%)accuracy: 27.10 -> 29.11Improved
use BiaSwap instead of LfFLfF · bFFHQ bias-guiding testaccuracy: 59.85 -> 99.13Improved
use BiaSwap instead of ReBias on real dataReBias · BAR bias-contrary testaccuracy: 39.29 -> 52.4Improved
use BiaSwap instead of ReBiasReBias · Colored MNIST unbiased (95.0%)accuracy: 96.28 -> 90.85Hurt
use BiaSwap instead of vanillaclassifier · BAR bias-guiding testaccuracy: 95.00 -> 93.33Hurt

Later papers that compared with this method

2 later accepted papers report BiaSwap as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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