mlsanityData to 14 Oct 2026

Improved Algorithm for Deep Active Learning under Imbalance via Optimal Separation

arXiv 2312.09196 ↗First posted 14 Dec 2023cs.LG cs.AI cs.CVcited by 8

Not on the map. Why?

Abstract (arXiv)

Class imbalance severely impacts machine learning performance on minority classes in real-world applications. While various solutions exist, active learning offers a fundamental fix by strategically collecting balanced, informative labeled examples from abundant unlabeled data. We introduce DIRECT, an algorithm that identifies class separation boundaries and selects the most uncertain nearby examples for annotation. By reducing the problem to one-dimensional active learning, DIRECT leverages established theory to handle batch labeling and label noise -- another common challenge in data annotation that particularly affects active learning methods. Our work presents the first comprehensive study of active learning under both class imbalance and label noise. Extensive experiments on imbalanced datasets show DIRECT reduces annotation costs by over 60% compared to state-of-the-art active learning methods and over 80% versus random sampling, while maintaining robustness to label noise.

Key facts. Cited by 8 papers in the map (data through 14 Oct 2026). Builds on 29 of its 52 references that are in the map. First posted on arXiv 14 Dec 2023.

Builds on

29 of 52 references.

References · 29

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2019When Does Label Smoothing Help?cited by 988
  4. 2017Deep Bayesian Active Learning with Image Datacited by 819
  5. 2021MedMNIST v2 -- A large-scale lightweight benchmark for 2D and 3D biomedical image classificationcited by 690
  6. 2019Deep Batch Active Learning by Diverse, Uncertain Gradient Lower Boundscited by 610
  7. 2021FlexMatch: Boosting Semi-Supervised Learning with Curriculum Pseudo Labelingcited by 540
  8. 2017Functional Map of the Worldcited by 338
  9. 2018Adversarial Active Learning for Deep Networks: a Margin Based Approachcited by 158
  10. 2021Batch Active Learning at Scalecited by 127
  11. 2021Gone Fishing: Neural Active Learning with Fisher Embeddingscited by 88
  12. 2017Deep Active Learning over the Long Tailcited by 84
17 more
  1. 2021SIMILAR: Submodular Information Measures Based Active Learning In Realistic Scenarioscited by 84
  2. 2021The iWildCam 2021 Competition Datasetcited by 77
  3. 2020An Empirical Process Approach to the Union Bound: Practical Algorithms for Combinatorial and Linear Banditscited by 45
  4. 2021Deep Active Learning by Leveraging Training Dynamicscited by 32
  5. 2022GALAXY: Graph-based Active Learning at the Extremecited by 29
  6. 2021Active Learning at the ImageNet Scalecited by 28
  7. 2022Making Look-Ahead Active Learning Strategies Feasible with Neural Tangent Kernelscited by 25
  8. 2020Similarity Search for Efficient Active Learning and Search of Rare Conceptscited by 24
  9. 2022A Lagrangian Duality Approach to Active Learningcited by 24
  10. 2015Active Learning from Weak and Strong Labelerscited by 22
  11. 2023LabelBench: A Comprehensive Framework for Benchmarking Adaptive Label-Efficient Learningcited by 20
  12. 2023Algorithm Selection for Deep Active Learning with Imbalanced Datasetscited by 18
  13. 2015Efficient and Parsimonious Agnostic Active Learningcited by 17
  14. 2021Improved Algorithms for Agnostic Pool-based Active Classificationcited by 15
  15. 2020A New Perspective on Pool-Based Active Classification and False-Discovery Controlcited by 9
  16. 2022Neural Active Learning on Heteroskedastic Distributionscited by 5
  17. 2022Improved Adaptive Algorithm for Scalable Active Learning with Weak Labelercited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 3; 2025: 2; 2026: 3.

8 papers.

  1. 2024An Experimental Design Framework for Label-Efficient Supervised Finetuning of Large Language Modelscited by 23
  2. 2024AHA: Human-Assisted Out-of-Distribution Generalization and Detectioncited by 7
  3. 2024Deep Active Learning in the Open Worldcited by 5
  4. 2025Active Learning with a Noisy Annotatorcited by 3
  5. 2025Unique Rashomon Sets for Robust Active Learningcited by 2
  6. 2026Active Learning with Foundation Model Priors: Efficient Learning under Class Imbalancecited by 0
  7. 2026Decoding Insect Song: A Multitask Semisupervised Orthoptera Bioacoustic Classifiercited by 0
  8. 2026Hard Cases, Bad Labels: Testing Error Exposure and Error Location in Uncertainty Sampling Under Bounded Label Noisecited by 0