Paper Draft: HAT Batch Normalization

This work aims to adapt batch normalization to be compatible with HAT architecture, which improves the absolute performance of HAT-based continual learning approaches.
research
Published

May 29, 2025