Rethinking deep active learning: Using unlabeled data at model training

2019/11/19 by Oriane Siméoni, Siméoni, Oriane, Mateusz Budnik +5 · 6 citations
Computer Science · Materials Science · #Computer Vision and Pattern Recognition (cs.CV) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning and Algorithms #Machine Learning and Data Classification #Machine Learning in Materials Science

paper · pdf · doi:10.48550/arxiv.1911.08177

openalex publication_date 2019/11/19 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/29

Abstract

Active learning typically focuses on training a model on few labeled examples alone, while unlabeled ones are only used for acquisition. In this work we depart from this setting by using both labeled and unlabeled data during model training across active learning cycles. We do so by using unsupervised feature learning at the beginning of the active learning pipeline and semi-supervised learning at every active learning cycle, on all available data. The former has not been investigated before in active learning, while the study of latter in the context of deep learning is scarce and recent findings are not conclusive with respect to its benefit. Our idea is orthogonal to acquisition strategies by using more data, much like ensemble methods use more models. By systematically evaluating on a number of popular acquisition strategies and datasets, we find that the use of unlabeled data during model training brings a surprising accuracy improvement in image classification, compared to the differences between acquisition strategies. We thus explore smaller label budgets, even one label per class.

Citations

• Gradient-based learning applied to document recognition
• Biased Importance Sampling for Deep Neural Network Training
• Active Learning for Convolutional Neural Networks: A Core-Set Approach
• Adversarial Active Learning for Deep Networks: a Margin Based Approach
• MixMatch: A Holistic Approach to Semi-Supervised Learning
• Deep Clustering for Unsupervised Learning of Visual Features
• Deep Bayesian Active Learning with Image Data
• Temporal Ensembling for Semi-Supervised Learning
• Learning Multiple Layers of Features from Tiny Images

Cited by

• Active Learning for GCN-based Action Recognition
• Label-Efficient Skeleton-based Recognition with Stable-Invertible Graph Convolutional Networks
• Cold-Start Active Preference Learning in Socio-Economic Domains
• Exploring Active Learning for Semiconductor Defect Segmentation
• A Novel Active Learning Approach to Label One Million Unknown Malware Variants
• A survey of human-in-the-loop for machine learning

Related

• Consistency-based Semi-supervised Active Learning: Towards Minimizing Labeling Cost
• Effective Evaluation of Deep Active Learning on Image Classification Tasks
• Parting with Illusions about Deep Active Learning
• Active Learning by Feature Mixing
• Reducing Label Effort: Self-Supervised meets Active Learning