Divide and not forget: Ensemble of selectively trained experts in Continual Learning
The article presents a novel approach in class-incremental learning named SEED. It addresses the issue of models forgetting previously acquired knowledge by selecting the most optimal expert for a task…
Continue reading