AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

A Simple Data Mixing Prior for Improving Self-Supervised Learning

Sucheng Ren, Huiyu Wang, Zhengqi Gao, Shengfeng He, Alan Yuille, Yuyin Zhou, Cihang Xie

CVPR 2022 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Invariance-based self-supervised learning · 36 citations (OpenAlex)

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 SDMP, treating source-to-mix and related mixed images as weighted positive pairsMoCo v3 with ViT-S · ImageNet linear evaluationTop-1 accuracy (%): 73.2% to 73.8%Improved
Add SDMPDINO with ViT-S · ImageNet linear evaluationTop-1 accuracy (%): 76.0% to 76.4%Improved
Add SDMPMoCo v3 with ResNet-50 · ImageNet linear evaluationTop-1 accuracy (%): 72.8% to 73.5%Improved
Add SDMPMoCo v3 with ViT-S · ImageNet end-to-end finetuningTop-1 accuracy: 78.7 to 79.1Improved
Add SDMPMoCo v3 with ViT-S · ImageNet semi-supervised, 1% labelsTop-1 accuracy: 54.4 to 55.5Improved
Add SDMPDINO with ViT-S · ImageNet-ATop-1 accuracy (%): 20.0 -> 21.1Improved
Add SDMPMoCo v3 with ResNet-50 · CIFAR-10 linear evaluation, 200 epochsAccuracy: 86.5 -> 89.5Improved
Add SDMPMoCo v3 with ResNet-50 · CIFAR-100 linear evaluation, 200 epochsAccuracy: 63.2 -> 68.2Improved
Replace sampled source-loss mixing weights with a static weight (ablation)SDMP with MoCo v3 · ImageNet linear evaluationTop-1 accuracy: 73.8 to 71.5Hurt
Replace sampled mixing-loss weights with a static weight (ablation)SDMP with MoCo v3 · ImageNet linear evaluationTop-1 accuracy: 73.8 to 72.5Hurt
Replace independent per-sample mixing coefficients with shared per-batch coefficients (ablation)SDMP with MoCo v3 · ImageNet linear evaluationTop-1 accuracy: 73.8 to 72.3Hurt
Use Mixup alone rather than selecting among Mixup, CutMix, and ResizeMix (ablation)SDMP with MoCo v3 · ImageNet linear evaluation, 100 epochsTop-1 accuracy (%): 65.5 to 65.1Hurt

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