Self-labelling via simultaneous clustering and representation learning

Abstract

Combining clustering and representation learning is one of the most promising approaches for unsupervised learning of deep neural networks. However, doing so naively leads to ill posed learning problems with degenerate solutions. In this paper, we propose a novel and principled learning formulation that addresses these issues. The method is obtained by maximizing the information between labels and input data indices. We show that this criterion extends standard crossentropy minimization to an optimal transport problem, which we solve efficiently for millions of input images and thousands of labels using a fast variant of the Sinkhorn-Knopp algorithm. The resulting method is able to self-label visual data so as to train highly competitive image representations without manual labels. Our method achieves state of the art representation learning performance for AlexNet and ResNet-50 on SVHN, CIFAR-10, CIFAR-100 and ImageNet and yields the first self-supervised AlexNet that outperforms the supervised Pascal VOC detection baseline. Code and models are available.

Accepted paper at the International Conference on Learning Representations (ICLR) 2020

References in corpus (4)

• One weird trick for parallelizing convolutional neural networks
• Learning Representations by Maximizing Mutual Information Across Views
• Unsupervised Deep Learning by Neighbourhood Discovery
• Revisiting Self-Supervised Visual Representation Learning

Cited by in corpus (15)

• Twin Contrastive Learning for Online Clustering
• One Million Scenes for Autonomous Driving: ONCE Dataset
• Deep Robust Clustering by Contrastive Learning
• Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation
• Conditional Negative Sampling for Contrastive Learning of Visual Representations
• A Prototype-Oriented Framework for Unsupervised Domain Adaptation
• AdCo: Adversarial Contrast for Efficient Learning of Unsupervised Representations from Self-Trained Negative Adversaries
• What is Wrong with One-Class Anomaly Detection?
• A Theory-Driven Self-Labeling Refinement Method for Contrastive Representation Learning
• Deep Contrastive Graph Representation via Adaptive Homotopy Learning
• Loss Function Entropy Regularization for Diverse Decision Boundaries
• Improving Image Clustering through Sample Ranking and Its Application to remote--sensing images
• Unsupervised Object Detection with LiDAR Clues
• Progressive Stage-wise Learning for Unsupervised Feature Representation Enhancement
• UniMoCo: Unsupervised, Semi-Supervised and Full-Supervised Visual Representation Learning