Learning to Self-Train for Semi-Supervised Few-Shot Classification

Neural Information Processing Systems 

Few-shot classification (FSC) is challenging due to the scarcity of labeled training data (e.g.