mlsanityData to 14 Oct 2026

Maintaining Plasticity in Continual Learning via Regenerative Regularization

arXiv 2308.11958 ↗First posted 23 Aug 2023cs.LG cs.AIcited by 65

Not on the map. Why?

Cited by 22 in Deep reinforcement learning methods, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

In continual learning, plasticity refers to the ability of an agent to quickly adapt to new information. Neural networks are known to lose plasticity when processing non-stationary data streams. In this paper, we propose L2 Init, a simple approach for maintaining plasticity by incorporating in the loss function L2 regularization toward initial parameters. This is very similar to standard L2 regularization (L2), the only difference being that L2 regularizes toward the origin. L2 Init is simple to implement and requires selecting only a single hyper-parameter. The motivation for this method is the same as that of methods that reset neurons or parameter values. Intuitively, when recent losses are insensitive to particular parameters, these parameters should drift toward their initial values. This prepares parameters to adapt quickly to new tasks. On problems representative of different types of nonstationarity in continual supervised learning, we demonstrate that L2 Init most consistently mitigates plasticity loss compared to previously proposed approaches.

Key facts. Cited by 65 papers in the map (data through 14 Oct 2026). Builds on 24 of its 30 references that are in the map. First posted on arXiv 23 Aug 2023.

Builds on

24 of 30 references.

References · 24

  1. 2016Layer Normalizationcited by 5,669
  2. 2016Overcoming catastrophic forgetting in neural networkscited by 4,787
  3. 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,611
  4. 2013Playing Atari with Deep Reinforcement Learningcited by 3,274
  5. 2017Gradient Episodic Memory for Continual Learningcited by 1,898
  6. 2018Riemannian Walk for Incremental Learning: Understanding Forgetting and Intransigencecited by 742
  7. 2013An Empirical Investigation of Catastrophic Forgetting in Gradient-Based Neural Networkscited by 722
  8. 2022The Primacy Bias in Deep Reinforcement Learningcited by 247
  9. 2023The Dormant Neuron Phenomenon in Deep Reinforcement Learningcited by 174
  10. 2023Understanding plasticity in neural networkscited by 174
  11. 2020Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learningcited by 134
  12. 2022Understanding and Preventing Capacity Loss in Reinforcement Learningcited by 125
12 more
  1. 2023Loss of Plasticity in Continual Deep Reinforcement Learningcited by 121
  2. 2016Understanding and Improving Convolutional Neural Networks via Concatenated Rectified Linear Unitscited by 86
  3. 2021Continual Backprop: Stochastic Gradient Descent with Persistent Randomnesscited by 80
  4. 2017Critical Learning Periods in Deep Neural Networkscited by 79
  5. 2020Transient Non-Stationarity and Generalisation in Deep Reinforcement Learningcited by 75
  6. 2023Deep Reinforcement Learning with Plasticity Injectioncited by 74
  7. 2021Online Continual Learning with Natural Distribution Shifts: An Empirical Study with Visual Datacited by 69
  8. 2019Bad Global Minima Exist and SGD Can Reach Themcited by 49
  9. 2023Maintaining Plasticity in Deep Continual Learningcited by 43
  10. 2023Online Continual Learning Without the Storage Constraintcited by 29
  11. 2022An Empirical Study of Implicit Regularization in Deep Offline RLcited by 24
  12. 2021Does the Adam Optimizer Exacerbate Catastrophic Forgetting?cited by 5

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 6; 2025: 27; 2026: 32.

65 papers.

Sort:

By area:

  1. 2024Overestimation, Overfitting, and Plasticity in Actor-Critic: the Bitter Lesson of Reinforcement Learningcited by 50
  2. 2024No Representation, No Trust: Connecting Representation, Collapse, and Trust Issues in PPOcited by 27
  3. 2025SFT Doesn't Always Hurt General Capabilities: Revisiting Domain-Specific Fine-Tuning in LLMscited by 24
  4. 2025Mitigating Plasticity Loss in Continual Reinforcement Learning by Reducing Churncited by 22
  5. 2024Non-Stationary Learning of Neural Networks with Automatic Soft Parameter Resetcited by 17
  6. 2025Optimizers Qualitatively Alter Solutions And We Should Leverage Thiscited by 17
  7. 2025Measure gradients, not activations! Enhancing neuronal activity in deep reinforcement learningcited by 14
  8. 2025Rethinking Safety in LLM Fine-tuning: An Optimization Perspectivecited by 14
  9. 2025Simplicial Embeddings Improve Sample Efficiency in Actor-Critic Agentscited by 12
  10. 2024Position: Lifetime tuning is incompatible with continual reinforcement learningcited by 11
  11. 2025A Forget-and-Grow Strategy for Deep Reinforcement Learning Scaling in Continuous Controlcited by 9
  12. 2026FIRE: Frobenius-Isometry Reinitialization for Balancing the Stability-Plasticity Tradeoffcited by 7
38 more
  1. 2025Reinitializing weights vs units for maintaining plasticity in neural networkscited by 6
  2. 2025Activation Function Design Sustains Plasticity in Continual Learningcited by 6
  3. 2025Context-Free Synthetic Data Mitigates Forgettingcited by 5
  4. 2025Barriers for Learning in an Evolving World: Mathematical Understanding of Loss of Plasticitycited by 5
  5. 2025Same accuracy, twice as fast: continuous training surpasses retraining from scratchcited by 4
  6. 2025Variance Control via Weight Rescaling in LLM Pre-trainingcited by 4
  7. 2025Experience Replay Addresses Loss of Plasticity in Continual Learningcited by 4
  8. 2025Optimistic critics can empower small actorscited by 4
  9. 2026Non-Uniform Noise-to-Signal Ratio in the REINFORCE Policy-Gradient Estimatorcited by 4
  10. 2026Weight Decay Improves Language Model Plasticitycited by 4
  11. 2026Predicting Plasticity in Deep Continual Learning: A Theoretical Perspectivecited by 4
  12. 2026Rotation-Preserving Supervised Fine-Tuningcited by 3
  13. 2024Hierarchical Subspaces of Policies for Continual Offline Reinforcement Learningcited by 2
  14. 2024Torque-Aware Momentumcited by 2
  15. 2025Overcoming Non-stationary Dynamics with Evidential Proximal Policy Optimizationcited by 2
  16. 2025Evaluating Temporal Plasticity in Foundation Time Series Models for Incremental Fine-tuningcited by 2
  17. 2025Plasticine: Accelerating Research in Plasticity-Motivated Deep Reinforcement Learningcited by 2
  18. 2025Preserving Plasticity in Continual Learning with Adaptive Linearity Injectioncited by 2
  19. 2025A Continual Offline Reinforcement Learning Benchmark for Navigation Taskscited by 2
  20. 2026Do Neural Networks Lose Plasticity in a Gradually Changing World?cited by 2
  21. 2025Silent Neuron Theory and Plasticity Preservation for Deep Reinforcement Learning in Adaptive Video Streamingcited by 1
  22. 2025Revisiting Clustering of Neural Bandits: Selective Reinitialization for Mitigating Loss of Plasticitycited by 1
  23. 2025On Understanding of the Dynamics of Model Capacity in Continual Learningcited by 1
  24. 2025An exact multiple-time-step variational formulation for the committor and the transition ratecited by 1
  25. 2025Intrinsic preservation of plasticity in continual quantum learningcited by 1
  26. 2026Beyond Single-Model Optimization: Preserving Plasticity in Continual Reinforcement Learningcited by 1
  27. 2026Recovery Guarantees for Continual Learning of Dependent Tasks: Memory, Data-Dependent Regularization, and Data-Dependent Weightscited by 1
  28. 2026Learning to Forget: Continual Learning with Adaptive Weight Decaycited by 1
  29. 2026Preserving Plasticity in Continual Learning via Dynamical Isometrycited by 1
  30. 2026SFT Overtraining Predicts Rank Inversion via Entropy Collapse Under RLVRcited by 1
  31. 2026Can Scale Save Us From Plasticity Loss in Large Language Models?cited by 1
  32. 2026Position: Collusion Risks Among AI Reasoning Agents Justify Certification Requirements for Making Market Decisionscited by 1
  33. 2025Neuron-level Balance between Stability and Plasticity in Deep Reinforcement Learningcited by 0
  34. 2025Class-Incremental Learning for Honey Botanical Origin Classification with Hyperspectral Images: A Study with Continual Backpropagationcited by 0
  35. 2026PLATE: Plasticity-Tunable Efficient Adapters for Geometry-Aware Continual Learningcited by 0
  36. 2026Continual Learning for non-stationary regression via Memory-Efficient Replaycited by 0
  37. 2026Rethinking Plasticity in Deep Reinforcement Learningcited by 0
  38. 2026Regularizing Attention Scores with Bootstrappingcited by 0

50 most cited of 65 All 65 (JSON)