Unsupervised Deep Learning by Neighbourhood Discovery

Abstract

Deep convolutional neural networks (CNNs) have demonstrated remarkable success in computer vision by supervisedly learning strong visual feature representations. However, training CNNs relies heavily on the availability of exhaustive training data annotations, limiting significantly their deployment and scalability in many application scenarios. In this work, we introduce a generic unsupervised deep learning approach to training deep models without the need for any manual label supervision. Specifically, we progressively discover sample anchored/centred neighbourhoods to reason and learn the underlying class decision boundaries iteratively and accumulatively. Every single neighbourhood is specially formulated so that all the member samples can share the same unseen class labels at high probability for facilitating the extraction of class discriminative feature representations during training. Experiments on image classification show the performance advantages of the proposed method over the state-of-the-art unsupervised learning models on six benchmarks including both coarse-grained and fine-grained object image categorisation.

36th International Conference on Machine Learning (ICML'19). Code is available at https://github.com/Raymond-sci/AND

Cited by in corpus (20)

• Unsupervised Learning of Visual Features by Contrasting Cluster Assignments
• Universal Domain Adaptation through Self Supervision
• Self-labelling via simultaneous clustering and representation learning
• Revisiting Contrastive Methods for Unsupervised Learning of Visual Representations
• Understanding the Behaviour of Contrastive Loss
• Unsupervised Representation Learning by InvariancePropagation
• Dual-Refinement: Joint Label and Feature Refinement for Unsupervised Domain Adaptive Person Re-Identification
• Unsupervised Multi-Source Domain Adaptation for Person Re-Identification
• Self-Adaptive Training: Bridging Supervised and Self-Supervised Learning
• Calibrating and Improving Graph Contrastive Learning
• Progressive Cluster Purification for Unsupervised Feature Learning
• Unsupervised Transfer Learning with Self-Supervised Remedy
• Dense Semantic Contrast for Self-Supervised Visual Representation Learning
• Clustering-Based Representation Learning through Output Translation and Its Application to Remote--Sensing Images
• A Comprehensive Approach to Unsupervised Embedding Learning based on AND Algorithm
• Unsupervised Image Classification for Deep Representation Learning
• Improving Image Clustering through Sample Ranking and Its Application to remote--sensing images
• Reliable Label Bootstrapping for Semi-Supervised Learning
• Unsupervised Embedding Learning from Uncertainty Momentum Modeling
• TraND: Transferable Neighborhood Discovery for Unsupervised Cross-domain Gait Recognition