A Simple Framework for Contrastive Learning of Visual Representations

13 Feb 2020Ting ChenSimon KornblithMohammad NorouziGeoffrey Hinton

This paper presents SimCLR: a simple framework for contrastive learning of visual representations. We simplify recently proposed contrastive self-supervised learning algorithms without requiring specialized architectures or a memory bank... (read more)

PDF Abstract
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT LEADERBOARD
Self-Supervised Image Classification ImageNet SimCLR (ResNet-50 4x) Top 1 Accuracy 76.5% # 6
Top 5 Accuracy 93.2% # 5
Number of Params 375M # 3
Self-Supervised Image Classification ImageNet SimCLR (ResNet-50 2x) Top 1 Accuracy 74.2% # 11
Top 5 Accuracy 92.0% # 7
Number of Params 94M # 8
Self-Supervised Image Classification ImageNet SimCLR (ResNet-50) Top 1 Accuracy 69.3% # 18
Top 5 Accuracy 89.0% # 13
Number of Params 24M # 11
Semi-Supervised Image Classification ImageNet - 10% labeled data SimCLR (ResNet-50 2×) Top 5 Accuracy 91.2% # 8
Semi-Supervised Image Classification ImageNet - 10% labeled data SimCLR (ResNet-50) Top 5 Accuracy 87.8% # 13
Semi-Supervised Image Classification ImageNet - 10% labeled data SimCLR (ResNet-50 4×) Top 5 Accuracy 92.6% # 4
Semi-Supervised Image Classification ImageNet - 1% labeled data SimCLR (ResNet-50 4×) Top 5 Accuracy 85.8% # 8
Top 1 Accuracy 63.0% # 8
Semi-Supervised Image Classification ImageNet - 1% labeled data SimCLR (ResNet-50 2×) Top 5 Accuracy 83.0% # 10
Top 1 Accuracy 58.5% # 10
Semi-Supervised Image Classification ImageNet - 1% labeled data SimCLR (ResNet-50) Top 5 Accuracy 75.5% # 16
Top 1 Accuracy 48.3% # 15

Methods used in the Paper