Data-Efficient Image Recognition with Contrastive Predictive Coding

Human observers can learn to recognize new categories of images from a handful of examples, yet doing so with artificial ones remains an open challenge. We hypothesize that data-efficient recognition is enabled by representations which make the variability in natural signals more predictable. We therefore revisit and improve Contrastive Predictive Coding, an unsupervised objective for learning such representations. This new implementation produces features which support state-of-the-art linear classification accuracy on the ImageNet dataset. When used as input for non-linear classification with deep neural networks, this representation allows us to use 2-5x less labels than classifiers trained directly on image pixels. Finally, this unsupervised representation substantially improves transfer learning to object detection on the PASCAL VOC dataset, surpassing fully supervised pre-trained ImageNet classifiers.

PDF Abstract ICML 2020 PDF


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Self-Supervised Image Classification ImageNet CPC v2 (ResNet-50) Top 1 Accuracy 63.8% # 97
Top 5 Accuracy 85.3% # 29
Number of Params 24M # 40
Self-Supervised Image Classification ImageNet CPC v2 (ResNet-161) Top 1 Accuracy 71.5% # 82
Top 5 Accuracy 90.1 # 23
Number of Params 305M # 19
Semi-Supervised Image Classification ImageNet - 10% labeled data CPC v2 (ResNet-161) Top 5 Accuracy 91.2% # 14
Top 1 Accuracy 73.1% # 26
Contrastive Learning imagenet-1k ResNet50 (v2) ImageNet Top-1 Accuracy 63.8 # 6
Contrastive Learning imagenet-1k ResNet v2 101 ImageNet Top-1 Accuracy 48.7 # 14