Optimal protocols for continual learning via statistical physics and control theory*
Authors:
Open Access
- 1 August 2025
- journal article
- Published by IOP Publishing in Journal of Statistical Mechanics: Theory and Experiment
- Vol. 2025 (8) , 084004
Abstract
Artificial neural networks often struggle with catastrophic forgetting when learning multiple tasks sequentially, as training on new tasks degrades performance on previously learned tasks. Recent theoretical work has addressed this issue by analysing learning curves in synthetic frameworks under predefined training protocols. However, these protocols rely on heuristics and lack a solid theoretical foundation for assessing their optimality. In this paper, we fill this gap by combining exact equations for training dynamics, derived using statistical physics techniques, with optimal control methods. We apply this approach to teacher–student models for continual learning and multi-task problems, obtaining a theory for task-selection protocols that maximises performance while minimising forgetting. Our theoretical analysis offers nontrivial yet interpretable strategies for mitigating catastrophic forgetting, shedding light on how optimal learning protocols modulate established effects, such as the influence of task similarity on forgetting. Finally, we validate our theoretical findings with experiments on real-world data.Keywords
All Related Versions
This publication has 29 references indexed in Scilit:
- Trainability and Accuracy of Artificial Neural Networks: An Interacting Particle System ApproachCommunications on Pure and Applied Mathematics, 2022
- Continual lifelong learning with neural networks: A reviewNeural Networks, 2019
- A mean-field optimal control formulation of deep learningResearch in the Mathematical Sciences, 2018
- A mean field view of the landscape of two-layer neural networksProceedings of the National Academy of Sciences, 2018
- Overcoming catastrophic forgetting in neural networksProceedings of the National Academy of Sciences of the United States of America, 2017
- Catastrophic forgetting in connectionist networksTrends in Cognitive Sciences, 1999
- Multitask LearningMachine Learning, 1997
- Semi-distributed Representations and Catastrophic Forgetting in Connectionist NetworksConnection Science, 1992
- Connectionist models of recognition memory: Constraints imposed by learning and forgetting functions.Psychological Review, 1990
- Three unfinished works on the optimal storage capacity of networksJournal of Physics A: General Physics, 1989