Predicting the Susceptibility of Examples to Catastrophic Forgetting
This work addresses the problem of catastrophic forgetting for continual learning practitioners, offering an incremental improvement by optimizing replay buffer composition.
The paper tackled catastrophic forgetting in neural networks by linking learning speed to forgetting susceptibility, showing that quickly learned examples are less prone to forgetting. It introduced Speed-Based Sampling (SBS), a strategy that selects replay examples based on learning speed, which improved performance across continual learning benchmarks and advanced state-of-the-art results.
Catastrophic forgetting - the tendency of neural networks to forget previously learned data when learning new information - remains a central challenge in continual learning. In this work, we adopt a behavioral approach, observing a connection between learning speed and forgetting: examples learned more quickly are less prone to forgetting. Focusing on replay-based continual learning, we show that the composition of the replay buffer - specifically, whether it contains quickly or slowly learned examples - has a significant effect on forgetting. Motivated by this insight, we introduce Speed-Based Sampling (SBS), a simple yet general strategy that selects replay examples based on their learning speed. SBS integrates easily into existing buffer-based methods and improves performance across a wide range of competitive continual learning benchmarks, advancing state-of-the-art results. Our findings underscore the value of accounting for the forgetting dynamics when designing continual learning algorithms.