A Simple Framework for Contrastive Learning of Visual Representations

TL;DR

SimCLR introduces a straightforward contrastive learning framework that significantly improves visual representation quality without complex architectures, leveraging data augmentation, nonlinear transformations, and large batch training.

Contribution

The paper simplifies contrastive self-supervised learning, systematically studies key components, and achieves state-of-the-art results on ImageNet with a simple framework.

Findings

Data augmentation composition is crucial for effective learning.

Learnable nonlinear transformation enhances representation quality.

Larger batch sizes and more training steps improve contrastive learning.

Abstract

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. In order to understand what enables the contrastive prediction tasks to learn useful representations, we systematically study the major components of our framework. We show that (1) composition of data augmentations plays a critical role in defining effective predictive tasks, (2) introducing a learnable nonlinear transformation between the representation and the contrastive loss substantially improves the quality of the learned representations, and (3) contrastive learning benefits from larger batch sizes and more training steps compared to supervised learning. By combining these findings, we are able to considerably outperform previous methods for self-supervised and semi-supervised learning on ImageNet. A linear classifier trained on self-supervised representations learned by SimCLR achieves 76.5% top-1 accuracy, which is a 7% relative improvement over previous state-of-the-art, matching the performance of a supervised ResNet-50. When fine-tuned on only 1% of the labels, we achieve 85.8% top-5 accuracy, outperforming AlexNet with 100X fewer labels.