mlsanityData to 7 Oct 2026

Rehearsal revealed: The limits and merits of revisiting samples in continual learning

arXiv 2104.07446 ↗First posted 15 Apr 2021cs.LG cs.CVcited by 76

Not on the map. Why?

Abstract (arXiv)

Learning from non-stationary data streams and overcoming catastrophic forgetting still poses a serious challenge for machine learning research. Rather than aiming to improve state-of-the-art, in this work we provide insight into the limits and merits of rehearsal, one of continual learning's most established methods. We hypothesize that models trained sequentially with rehearsal tend to stay in the same low-loss region after a task has finished, but are at risk of overfitting on its sample memory, hence harming generalization. We provide both conceptual and strong empirical evidence on three benchmarks for both behaviors, bringing novel insights into the dynamics of rehearsal and continual learning in general. Finally, we interpret important continual learning works in the light of our findings, allowing for a deeper understanding of their successes.

Key facts. Cited by 76 papers in the map (data through 7 Oct 2026). Builds on 32 of its 46 references that are in the map. First posted on arXiv 15 Apr 2021.

Builds on

32 of 46 references.

References · 32

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2015Distilling the Knowledge in a Neural Networkcited by 10,076
  3. 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
  4. 2016Overcoming catastrophic forgetting in neural networkscited by 4,718
  5. 2016Learning without Forgettingcited by 2,465
  6. 2016iCaRL: Incremental Classifier and Representation Learningcited by 2,313
  7. 2016Matching Networks for One Shot Learningcited by 2,246
  8. 2017Gradient Episodic Memory for Continual Learningcited by 1,871
  9. 2016Understanding deep learning requires rethinking generalizationcited by 1,657
  10. 2016On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minimacited by 1,337
  11. 2016Progressive Neural Networkscited by 1,332
  12. 2018Averaging Weights Leads to Wider Optima and Better Generalizationcited by 1,194
20 more
  1. 2017Continual Learning with Deep Generative Replaycited by 1,152
  2. 2017Memory Aware Synapses: Learning what (not) to forgetcited by 1,094
  3. 2018Efficient Lifelong Learning with A-GEMcited by 1,044
  4. 2017PackNet: Adding Multiple Tasks to a Single Network by Iterative Pruningcited by 882
  5. 2018Overcoming catastrophic forgetting with hard attention to the taskcited by 672
  6. 2019Three scenarios for continual learningcited by 610
  7. 2018End-to-End Incremental Learningcited by 596
  8. 2019Gradient based sample selection for online continual learningcited by 568
  9. 2019Online Continual Learning with Maximally Interfered Retrievalcited by 376
  10. 2019Orthogonal Gradient Descent for Continual Learningcited by 361
  11. 2018Essentially No Barriers in Neural Network Energy Landscapecited by 328
  12. 2019On Tiny Episodic Memories in Continual Learningcited by 299
  13. 2018Re-evaluating Continual Learning Scenarios: A Categorization and Case for Strong Baselinescited by 228
  14. 2020Understanding the Role of Training Regimes in Continual Learningcited by 177
  15. 2020Linear Mode Connectivity in Multitask and Continual Learningcited by 144
  16. 2020Continual Prototype Evolution: Learning Online from Non-Stationary Data Streamscited by 143
  17. 2018Rotate your Networks: Better Weight Consolidation and Less Catastrophic Forgettingcited by 127
  18. 2021Avalanche: an End-to-End Library for Continual Learningcited by 124
  19. 2017Continual Learning in Generative Adversarial Netscited by 51
  20. 2020Unsupervised Model Personalization while Preserving Privacy and Scalability: An Open Problemcited by 13

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 4; 2022: 12; 2023: 21; 2024: 21; 2025: 9; 2026: 9.

76 papers.

Sort:

By area:

  1. 2023A Comprehensive Survey of Continual Learning: Theory, Method and Applicationcited by 860
  2. 2021DyTox: Transformers for Continual Learning with DYnamic TOken eXpansioncited by 291
  3. 2023Class-Incremental Learning: A Surveycited by 209
  4. 2022Robust Mean Teacher for Continual and Gradual Test-Time Adaptationcited by 102
  5. 2022Representation Compensation Networks for Continual Semantic Segmentationcited by 72
  6. 2022Continual evaluation for lifelong learning: Identifying the stability gapcited by 55
  7. 2023Federated Continual Learning via Knowledge Fusion: A Surveycited by 53
  8. 2022On the Effectiveness of Lipschitz-Driven Rehearsal in Continual Learningcited by 39
  9. 2023The Ideal Continual Learner: An Agent That Never Forgetscited by 34
  10. 2022A simple but strong baseline for online continual learning: Repeated Augmented Rehearsalcited by 32
  11. 2023Active Neural Mappingcited by 30
  12. 2024CLAP4CLIP: Continual Learning with Probabilistic Finetuning for Vision-Language Modelscited by 25
38 more
  1. 2023Catastrophic Forgetting in Deep Learning: A Comprehensive Taxonomycited by 24
  2. 2022Online Continual Learning with Contrastive Vision Transformercited by 21
  3. 2022Learning to Imagine: Diversify Memory for Incremental Learning using Unlabeled Datacited by 19
  4. 2025Reinforced Refinement with Self-Aware Expansion for End-to-End Autonomous Drivingcited by 19
  5. 2021Lifelong Unsupervised Domain Adaptive Person Re-identification with Coordinated Anti-forgetting and Adaptationcited by 18
  6. 2023Two Complementary Perspectives to Continual Learning: Ask Not Only What to Optimize, But Also Howcited by 16
  7. 2024Orchestrate Latent Expertise: Advancing Online Continual Learning with Multi-Level Supervision and Reverse Self-Distillationcited by 16
  8. 2023Exploring Continual Learning of Diffusion Modelscited by 15
  9. 2023DualHSIC: HSIC-Bottleneck and Alignment for Continual Learningcited by 15
  10. 2023Class Incremental Learning with Pre-trained Vision-Language Modelscited by 15
  11. 2021Saliency Guided Experience Packing for Replay in Continual Learningcited by 14
  12. 2023A multifidelity approach to continual learning for physical systemscited by 13
  13. 2022Learn the Time to Learn: Replay Scheduling in Continual Learningcited by 11
  14. 2022D3Former: Debiased Dual Distilled Transformer for Incremental Learningcited by 10
  15. 2022Memory Population in Continual Learning via Outlier Eliminationcited by 9
  16. 2023Improving Continual Relation Extraction by Distinguishing Analogous Semanticscited by 9
  17. 2024Class-incremental Learning for Time Series: Benchmark and Evaluationcited by 9
  18. 2022Learning "O" Helps for Learning More: Handling the Concealed Entity Problem for Class-incremental NERcited by 8
  19. 2024Integrating Present and Past in Unsupervised Continual Learningcited by 8
  20. 2024Active Fine-Tuning of Multi-Task Policiescited by 8
  21. 2023Continual Learning with Strong Experience Replaycited by 7
  22. 2024PASS++: A Dual Bias Reduction Framework for Non-Exemplar Class-Incremental Learningcited by 7
  23. 2024Early Preparation Pays Off: New Classifier Pre-tuning for Class Incremental Semantic Segmentationcited by 7
  24. 2026Position: Modular Memory is the Key to Continual Learning Agentscited by 7
  25. 2021On robustness of generative representations against catastrophic forgettingcited by 5
  26. 2022Re-examining Distillation For Continual Object Detectioncited by 5
  27. 2024Overcoming Catastrophic Forgetting in Tabular Data Classification: A Pseudorehearsal-based approachcited by 5
  28. 2024An Efficient Rehearsal Scheme for Catastrophic Forgetting Mitigation during Multi-stage Fine-tuningcited by 4
  29. 2024IMEX-Reg: Implicit-Explicit Regularization in the Function Space for Continual Learningcited by 4
  30. 2025Replay Can Provably Increase Forgettingcited by 4
  31. 2023Studying Generalization on Memory-Based Methods in Continual Learningcited by 3
  32. 2023Class-Incremental Mixture of Gaussians for Deep Continual Learningcited by 3
  33. 2023Coordinated Replay Sample Selection for Continual Federated Learningcited by 3
  34. 2024Overcoming Catastrophic Forgetting by Exemplar Selection in Task-oriented Dialogue Systemcited by 3
  35. 2024How to Leverage Predictive Uncertainty Estimates for Reducing Catastrophic Forgetting in Online Continual Learningcited by 3
  36. 2024Leveraging Hierarchical Taxonomies in Prompt-based Continual Learningcited by 3
  37. 2023Class-Incremental Learning of Plant and Disease Detection: Growing Branches with Knowledge Distillationcited by 2
  38. 2024Watch Your Step: Optimal Retrieval for Continual Learning at Scalecited by 2

50 most cited of 76 All 76 (JSON)