SimCLR通过简单框架提升视觉表征学习，无需特殊结构即可超越此前自监督方法。
A Simple Framework for Contrastive Learning of Visual Representations
- 采用数据增强组合与非线性变换提升对比学习效果。
- 自监督表征线性分类达76.5%准确率，比之前高7%。
- 适合追求高效无监督表征的科研与工程人员。
本文提出SimCLR：一种用于视觉表征对比学习的简洁框架。我们简化了近期提出的对比自监督学习算法，无需专用网络结构或内存池。为理解对比预测任务为何能学习有效表征，我们系统研究了框架中的关键组件。结果表明：（1）数据增强的组合在定义有效预测任务中起关键作用；（2）在表示与对比损失之间引入可学习的非线性变换显著提升表征质量；（3）对比学习相比有监督学习更受益于更大批次和更多训练步数。结合这些发现，我们在ImageNet上显著优于以往自监督与半监督方法。基于自监督学习的表示训练线性分类器达到76.5%的top-1准确率，相较之前最优方法相对提升7%，媲美监督训练的ResNet-50。在仅使用1%标签微调时，达到85.8% top-5准确率，超过使用100倍标签的AlexNet。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。