Pseudo-Labeling Based Practical Semi-Supervised Meta-Training for Few-Shot Learning

Most existing few-shot learning (FSL) methods require a large amount of labeled data in meta-training, which is a major limit. To reduce the requirement of labels, a semi-supervised meta-training setting has been proposed for FSL, which includes only a few labeled samples and numbers of unlabeled samples in base classes. However, existing methods under this setting require class-aware sample selection from the unlabeled set, which violates the assumption of unlabeled set. In this paper, we propose a practical semi-supervised meta-training setting with truly unlabeled data. Under the new setting, the performance of existing methods drops notably. To better utilize both the labeled and truly unlabeled data, we propose a simple and effective meta-training framework, called pseudo-labeling based on meta-learning (PLML). Firstly, we train a classifier via common semi-supervised learning (SSL) and use it to obtain the pseudo-labels of unlabeled data. Then we build few-shot tasks from labeled and pseudo-labeled data and run meta-learning over the constructed tasks to learn the FSL model. Surprisingly, through extensive experiments across two FSL datasets, we find that this simple meta-training framework effectively prevents the performance degradation of FSL under limited labeled data. Besides, benefiting from meta-training, the proposed method improves the classifiers learned by two representative SSL algorithms as well.