How We Define Continual Learning

Continual learning (also called lifelong or incremental learning)¹ is the ability of a machine learning model to keep learning from a stream of new data over time, acquiring new knowledge and skills after deployment without forgetting what it has already learned, a failure mode known as catastrophic forgetting.² Unlike the standard paradigm, where a model is trained once on a fixed dataset and then frozen, a continually learning system updates itself incrementally as new information arrives,³ adapting to each user and task rather than staying static.