DASH: Warm-Starting Neural Network Training in Stationary Settings without Loss of Plasticity
This addresses a practical issue for deploying neural networks with continuous data updates, though it is incremental as it builds on existing warm-starting methods.
The paper tackled the problem of neural networks losing plasticity when warm-started on stationary data, identifying noise memorization as the cause and proposing DASH to mitigate it, resulting in improved test accuracy and training efficiency on vision tasks.
Warm-starting neural network training by initializing networks with previously learned weights is appealing, as practical neural networks are often deployed under a continuous influx of new data. However, it often leads to loss of plasticity, where the network loses its ability to learn new information, resulting in worse generalization than training from scratch. This occurs even under stationary data distributions, and its underlying mechanism is poorly understood. We develop a framework emulating real-world neural network training and identify noise memorization as the primary cause of plasticity loss when warm-starting on stationary data. Motivated by this, we propose Direction-Aware SHrinking (DASH), a method aiming to mitigate plasticity loss by selectively forgetting memorized noise while preserving learned features. We validate our approach on vision tasks, demonstrating improvements in test accuracy and training efficiency.