Trust Region Continual Learning as an Implicit Meta-Learner

TL;DR

This paper introduces a hybrid trust region approach combining generative replay and Fisher-metric constraints, which acts as an implicit meta-learner to improve continual learning performance and retention.

Contribution

It presents a novel trust region continual learning method that implicitly functions as a meta-learner, enhancing task retention without explicit bilevel optimization.

Findings

Achieves superior final performance and retention in diffusion tasks

Recovers early-task performance faster than existing methods

Acts as an implicit meta-learner with MAML-style interpretation

Abstract

Continual learning aims to acquire tasks sequentially without catastrophic forgetting, yet standard strategies face a core tradeoff: regularization-based methods (e.g., EWC) can overconstrain updates when task optima are weakly overlapping, while replay-based methods can retain performance but drift due to imperfect replay. We study a hybrid perspective: \emph{trust region continual learning} that combines generative replay with a Fisher-metric trust region constraint. We show that, under local approximations, the resulting update admits a MAML-style interpretation with a single implicit inner step: replay supplies an old-task gradient signal (query-like), while the Fisher-weighted penalty provides an efficient offline curvature shaping (support-like). This yields an emergent meta-learning property in continual learning: the model becomes an initialization that rapidly \emph{re-converges} to prior task optima after each task transition, without explicitly optimizing a bilevel objective. Empirically, on task-incremental diffusion image generation and continual diffusion-policy control, trust region continual learning achieves the best final performance and retention, and consistently recovers early-task performance faster than EWC, replay, and continual meta-learning baselines.