Dynamic Feedback Engines: Layer-Wise Control for Self-Regulating Continual Learning

TL;DR

This paper introduces a dynamic, entropy-aware feedback mechanism for continual learning that adaptively regulates layer confidence to prevent overfitting and underfitting, improving performance on multiple datasets.

Contribution

It proposes a novel layer-wise entropy regulation method that enhances continual learning by balancing stability and plasticity, adaptable to existing approaches.

Findings

Significant performance improvements over state-of-the-art baselines.

Effective in reducing catastrophic forgetting across datasets.

Adaptive regulation leads to wider local minima and better generalization.

Abstract

Continual learning aims to acquire new tasks while preserving performance on previously learned ones, but most methods struggle with catastrophic forgetting. Existing approaches typically treat all layers uniformly, often trading stability for plasticity or vice versa. However, different layers naturally exhibit varying levels of uncertainty (entropy) when classifying tasks. High-entropy layers tend to underfit by failing to capture task-specific patterns, while low-entropy layers risk overfitting by becoming overly confident and specialized. To address this imbalance, we propose an entropy-aware continual learning method that employs a dynamic feedback mechanism to regulate each layer based on its entropy. Specifically, our approach reduces entropy in high-entropy layers to mitigate underfitting and increases entropy in overly confident layers to alleviate overfitting. This adaptive regulation encourages the model to converge to wider local minima, which have been shown to improve generalization. Our method is general and can be seamlessly integrated with both replay- and regularization-based approaches. Experiments on various datasets demonstrate substantial performance gains over state-of-the-art continual learning baselines.