Addressing Loss of Plasticity and Catastrophic Forgetting in Continual Learning

TL;DR

This paper introduces UPGD, a novel continual learning method that balances plasticity and stability by perturbing gradients based on unit usefulness, outperforming existing approaches in streaming learning and reinforcement learning tasks.

Contribution

The paper proposes Utility-based Perturbed Gradient Descent (UPGD), a new method that simultaneously addresses catastrophic forgetting and loss of plasticity in continual learning.

Findings

UPGD outperforms existing methods in streaming learning benchmarks.

UPGD maintains high accuracy over multiple tasks with non-stationarities.

In reinforcement learning, UPGD prevents performance drops seen with Adam.

Abstract

Deep representation learning methods struggle with continual learning, suffering from both catastrophic forgetting of useful units and loss of plasticity, often due to rigid and unuseful units. While many methods address these two issues separately, only a few currently deal with both simultaneously. In this paper, we introduce Utility-based Perturbed Gradient Descent (UPGD) as a novel approach for the continual learning of representations. UPGD combines gradient updates with perturbations, where it applies smaller modifications to more useful units, protecting them from forgetting, and larger modifications to less useful units, rejuvenating their plasticity. We use a challenging streaming learning setup where continual learning problems have hundreds of non-stationarities and unknown task boundaries. We show that many existing methods suffer from at least one of the issues, predominantly manifested by their decreasing accuracy over tasks. On the other hand, UPGD continues to improve performance and surpasses or is competitive with all methods in all problems. Finally, in extended reinforcement learning experiments with PPO, we show that while Adam exhibits a performance drop after initial learning, UPGD avoids it by addressing both continual learning issues.