Continual Learning

Learning continuously without forgetting past knowledge.

---

Continual learning enables models to learn over time while retaining previously acquired knowledge.

• Approaches: Regularization (e.g., EWC), rehearsal/replay, dynamic architectures.
• Challenges: Catastrophic forgetting, data availability, version-to-version validation.