Specifically, meta-learning is executed over a pretrained network for few-shot classification. Furthermore, self-supervised learning is exploited to fit the target domain distribution by training on ...

Deep learning has attracted wide attention recently because of its excellent feature representation ability and end-to-end automatic learning method. Especially in clinical medical imaging diagnosis, ...