Self-Distillation: Towards Efficient and Compact Neural Networks | IEEE Journals & Magazine | IEEE Xplore