Supervised Contrastive Learning

23 Apr 2020Prannay KhoslaPiotr TeterwakChen WangAaron SarnaYonglong TianPhillip IsolaAaron MaschinotCe LiuDilip Krishnan

Cross entropy is the most widely used loss function for supervised training of image classification models. In this paper, we propose a novel training methodology that consistently outperforms cross entropy on supervised learning tasks across different architectures and data augmentations... (read more)

PDF Abstract

Results from the Paper


#55 best model for Image Classification on ImageNet (using extra training data)

     Get a GitHub badge
TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK USES EXTRA
TRAINING DATA
RESULT LEADERBOARD
Image Classification ImageNet ResNet-200 (Supervised Contrastive) Top 1 Accuracy 80.8% # 55
Top 5 Accuracy 95.6% # 31

Methods used in the Paper