analiza aiqbench
This paper challenges the standard continual learning goal of retaining all past knowledge, arguing that in non-stationary settings, heavy retention can slow adaptation. It offers a formal decomposition (Transfer Efficiency, Critical Task Duration) and a proposed Predictive Continual Learning framework, with a Window algorithm shown only in a controlled drift setting. Readers should treat the theory and the proof-of-concept results as promising but narrow until they are tested on broader real-world benchmarks.
To Retain or to Adapt? Generalizing Continual Learning
arXiv:2607.05609v2 Announce Type: replace-cross Abstract: The Continual Learning (CL) literature has long been driven by the goal of mitigating catastrophic forgetting. This objective rests on an often unstated assumption: that a lifelong learner should approximate the Joint-Task Learning (JTL) solution and retain all previously acquired knowledge. We challenge this retention-centered premise, arguing that in non-stationary environments prioritizing retention can impede real-time adaptation. Shifting the focus to the Average Lifelong Error (ALE), we formalize CL as an online optimization problem governed by the interaction between environmental and learning dynamics. We introduce Transfer Efficiency as a quantitative measure of the tension between Instability, the bias inherited from conflicting past experience, and Transient Error, the optimization cost of learning new tasks from scratch. Under mild convergence conditions, holding across linear and neural network models, this decomposition yields a Critical Task Duration: a closed-form threshold beyond which historical knowledge transitions from a warm-start advantage to an optimization liability whenever retention induces a positive stationary bias. We validate these theoretical predictions on continual image classification and reinforcement learning benchmarks. Finally, by connecting continual learning to the online learning framework of predictable sequences, we show that JTL is only one instance of a broader family of objectives, and we propose a new general class of continual learning algorithms, which we call Predictive Continual Learning. Predictive CL algorithms optimize expected future performance under an explicit, dynamically updated model of future tasks. As a proof of concept, we analyze a Window algorithm that interpolates between JTL and Independent-Task Learning (ITL), outperforming both under controlled distributional drift.
Title:To Retain or to Adapt? Generalizing Continual Learning
Abstract:The Continual Learning (CL) literature has long been driven by the goal of mitigating catastrophic forgetting. This objective rests on an often unstated assumption: that a lifelong learner should approximate the Joint-Task Learning (JTL) solution and retain all previously acquired knowledge. We challenge this retention-centered premise, arguing that in non-stationary environments prioritizing retention can impede real-time adaptation. Shifting the focus to the Average Lifelong Error (ALE), we formalize CL as an online optimization problem governed by the interaction between environmental and learning dynamics. We introduce Transfer Efficiency as a quantitative measure of the tension between Instability, the bias inherited from conflicting past experience, and Transient Error, the optimization cost of learning new tasks from scratch. Under mild convergence conditions, holding across linear and neural network models, this decomposition yields a Critical Task Duration: a closed-form threshold beyond which historical knowledge transitions from a warm-start advantage to an optimization liability whenever retention induces a positive stationary bias. We validate these theoretical predictions on continual image classification and reinforcement learning benchmarks. Finally, by connecting continual learning to the online learning framework of predictable sequences, we show that JTL is only one instance of a broader family of objectives, and we propose a new general class of continual learning algorithms, which we call Predictive Continual Learning. Predictive CL algorithms optimize expected future performance under an explicit, dynamically updated model of future tasks. As a proof of concept, we analyze a Window algorithm that interpolates between JTL and Independent-Task Learning (ITL), outperforming both under controlled distributional drift.
Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2607.05609 [stat.ML]
(or arXiv:2607.05609v2 [stat.ML] for this version)
https://doi.org/10.48550/arXiv.2607.05609 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 359 cuvinte
Subiecte conexe
MePo++: Unifying Representation Refinement and Reconciliation for General Continual Learning
arXiv · 7 sept. 2026
What Drives Compositional Generalization in Visual Generative Models? The Importance of Continuous Training Objectives
arXiv · 2 oct. 2026
Episodic looking - using reinforcement learning to bias perception enhances few-shot and continual learning with protection from forgetting
arXiv · 31 oct. 2026
Dose-Aware Cold Diffusion with Physics Consistency for Generalizable Low-Dose CT Reconstruction
arXiv · 17 sept. 2026