AI Paper Atlas

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

Revisiting the Critical Factors of Augmentation-Invariant Representation Learning

Junqiang Huang, Xiangwen Kong, Xiangyu Zhang

ECCV 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 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 symmetric contrastive loss (ablation)MoCo v2 (with BN, predictor, momentum ascent) · ImageNet linear evaltop-1 accuracy (%): 2.0% accuracy incrementImproved
add all BYOL-style model configurations and complex augmentations (ablation)MoCo v2 · ImageNet linear eval (200 ep)top-1 accuracy (%): 72.0 -> 72.4Improved
add the BYOL-style configurations (MoCo v2+) (ablation)MoCo v2 · transfer (VOC, COCO)AP: about 0.5% improvementNo clear effect
pretrain with LARS instead of SGD (ablation)MoCo v2+ · VOC07 AP50AP50 (%): 77.1 -> 62.9Hurt
pretrain with LARS instead of SGD (ablation)BYOL · VOC07 AP50AP50 (%): 76.2 -> 71.7Hurt
add NormRescale to LARS-trained BYOLBYOL-LARS · VOC07 AP50AP50 (%): 71.7 -> 76.6Improved
use a constant 0.1 norm scale instead of SGD anchor in NormRescale (ablation)BYOL-NR · VOC07 AP50AP50 (%): 76.6 -> 76.1No clear effect
symmetrize the network (add teacher predictor) (ablation)MoCo v2+ · ImageNet linear eval (200 ep)top-1 accuracy (%): 72.4 -> 71.2Hurt
symmetrize the network (add teacher predictor) (ablation)MoCo v2+ · VOC07, COCO, CityScapes transferAP/mIoU: gap within 0.4No clear effect
symmetrize the network (add teacher predictor) (ablation)MoCo v2+ · CIFAR-10-LT and CIFAR-100-LTlinear and fine-tuning accuracyImproved
replace negative-free BYOL with contrastive loss (ablation)BYOL · CIFAR-10-LT and CIFAR-100-LTaccuracyImproved
remove data augmentations with color transforms (ablation)MoCo v2+, S-MoCo v2+, BYOL · ImageNet linear evaltop-1 accuracy (%)Hurt

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