Few-Shot and Continual Learning with Attentive Independent Mechanisms