PGADA: Perturbation-Guided Adversarial Alignment for Few-shot Learning Under the Support-Query Shift
This addresses a distribution shift problem in few-shot learning for real-world applications, but it is incremental as it builds on existing optimal transportation methods.
The paper tackles the problem of support-query shift in few-shot learning, where distribution shifts degrade model performance, and proposes PGADA with regularized optimal transportation, achieving significant outperformance over eleven state-of-the-art methods on three benchmark datasets.
Few-shot learning methods aim to embed the data to a low-dimensional embedding space and then classify the unseen query data to the seen support set. While these works assume that the support set and the query set lie in the same embedding space, a distribution shift usually occurs between the support set and the query set, i.e., the Support-Query Shift, in the real world. Though optimal transportation has shown convincing results in aligning different distributions, we find that the small perturbations in the images would significantly misguide the optimal transportation and thus degrade the model performance. To relieve the misalignment, we first propose a novel adversarial data augmentation method, namely Perturbation-Guided Adversarial Alignment (PGADA), which generates the hard examples in a self-supervised manner. In addition, we introduce Regularized Optimal Transportation to derive a smooth optimal transportation plan. Extensive experiments on three benchmark datasets manifest that our framework significantly outperforms the eleven state-of-the-art methods on three datasets.