mlsanityData to 7 Oct 2026

An Empirical Study on the Efficacy of Deep Active Learning for Image Classification

arXiv 2212.03088 ↗First posted 30 Nov 2022cs.CV cs.AIcited by 6

Not on the map. Why?

Abstract (arXiv)

Deep Active Learning (DAL) has been advocated as a promising method to reduce labeling costs in supervised learning. However, existing evaluations of DAL methods are based on different settings, and their results are controversial. To tackle this issue, this paper comprehensively evaluates 19 existing DAL methods in a uniform setting, including traditional fully-underlinesupervised underlineactive underlinelearning (SAL) strategies and emerging underlinesemi-underlinesupervised underlineactive underlinelearning (SSAL) techniques. We have several non-trivial findings. First, most SAL methods cannot achieve higher accuracy than random selection. Second, semi-supervised training brings significant performance improvement compared to pure SAL methods. Third, performing data selection in the SSAL setting can achieve a significant and consistent performance improvement, especially with abundant unlabeled data. Our findings produce the following guidance for practitioners: one should (i) apply SSAL early and (ii) collect more unlabeled data whenever possible, for better model performance.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Builds on 18 of its 31 references that are in the map. First posted on arXiv 30 Nov 2022.

Builds on

18 of 31 references.

References · 18

  1. 2019MixMatch: A Holistic Approach to Semi-Supervised Learningcited by 1,390
  2. 2017Deep Bayesian Active Learning with Image Datacited by 812
  3. 2019Deep Batch Active Learning by Diverse, Uncertain Gradient Lower Boundscited by 602
  4. 2019Pseudo-Labeling and Confirmation Bias in Deep Semi-Supervised Learningcited by 507
  5. 2019Learning Loss for Active Learningcited by 335
  6. 2019Variational Adversarial Active Learningcited by 284
  7. 2018Adversarial Active Learning for Deep Networks: a Margin Based Approachcited by 155
  8. 2021Batch Active Learning at Scalecited by 127
  9. 2019Consistency-based Semi-supervised Active Learning: Towards Minimizing Labeling Costcited by 114
  10. 2019Deep Active Learning: Unified and Principled Method for Query and Trainingcited by 92
  11. 2019Diverse mini-batch Active Learningcited by 88
  12. 2020Sequential Graph Convolutional Network for Active Learningcited by 69
6 more
  1. 2020Towards Robust and Reproducible Active Learning Using Neural Networkscited by 53
  2. 2018Adversarial Sampling for Active Learningcited by 51
  3. 2019Parting with Illusions about Deep Active Learningcited by 38
  4. 2021Effective Evaluation of Deep Active Learning on Image Classification Taskscited by 37
  5. 2020On the Marginal Benefit of Active Learning: Does Self-Supervision Eat Its Cake?cited by 18
  6. 2019Combining MixMatch and Active Learning for Better Accuracy with Fewer Labelscited by 17

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 2; 2024: 2; 2025: 1; 2026: 1.

6 papers.

  1. 2023ActiveGLAE: A Benchmark for Deep Active Learning with Transformerscited by 14
  2. 2024A Cross-Domain Benchmark for Active Learningcited by 11
  3. 2024GCI-ViTAL: Gradual Confidence Improvement with Vision Transformers for Active Learning on Label Noisecited by 2
  4. 2023Towards Comparable Active Learningcited by 0
  5. 2025Balancing Accuracy, Calibration, and Efficiency in Active Learning with Vision Transformers Under Label Noisecited by 0
  6. 2026Are Candidate Models Really Needed for Active Learning?cited by 0