AI Paper Atlas

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

Dual Temperature Helps Contrastive Learning Without Many Negative Samples: Towards Understanding and Simplifying MoCo

Chaoning Zhang, Kang Zhang, Trung X. Pham, Axi Niu, Zhinan Qiao, Chang D. Yoo, In So Kweon

CVPR 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Dense features and foundation priors · 43 citations (OpenAlex)

Check 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
Replace MoCo v2 with dictionary-free SimMoCo using dual temperatureMoCo v2 · CIFAR100Top-1 accuracy: 53.28 -> 54.11Improved
Remove momentum from SimMoCo to form SimCoSimMoCo · CIFAR100Top-1 accuracy: 54.11 -> 58.35Improved
Remove the dictionary and use dual temperatureMoCo v2 · STL10Top-1 accuracy: 81.25 -> 80.56Hurt
Remove the dictionary and momentum; use dual temperatureMoCo v2 · SVHNTop-1 accuracy: 47.25 -> 52.37Improved
Set the inter-anchor temperature higher than the intra-anchor temperature (ablation)SimMoCo · CIFAR100Top-1 accuracy: 49.52 -> 54.11Improved
Reverse the dual-temperature assignment, making τ α greater than τ β (ablation)SimMoCo · CIFAR100Top-1 accuracy: 54.11 -> 31.28Hurt
Use a very low scalar temperature, τ β=0.05, rather than τ β=0.5 (ablation)SimMoCo · CIFAR100Top-1 accuracy: 54.07 -> 43.56Hurt
Replace last-iteration vector keys with earliest dictionary keys (ablation)MoCo v2 vector component · CIFAR100Top-1 accuracy: 53.43 -> 1Hurt
Replace newest vector keys with randomly sampled keys (ablation)MoCo v2 vector component · CIFAR100Top-1 accuracy: 53.43 -> 44.03Hurt
Replace newest scalar keys with earliest scalar dictionary keys (ablation)MoCo v2 scalar component · CIFAR100Top-1 accuracy: 53.32 -> 52.13Hurt
Remove inter-anchor hardness-awareness (ablation)SimCo · CIFAR100Top-1 accuracy: 53.82 -> 58.35Improved
Remove inter-anchor hardness-awareness (ablation)SimMoCo · CIFAR100Top-1 accuracy: 50.22 -> 54.11Improved

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