论文部分内容阅读
A general framework to tackle the problem of few-shot learning is meta-learning, which aims to train a well-generalized meta-learner(or backbone network) to learn a base-learner for each future task with small training data. Although a lot of work has