PGADA: Perturbation-Guided Adversarial Alignment for Few-shot Learning Under the Support-Query Shift

TL;DR

This paper introduces PGADA, a novel adversarial data augmentation method that improves few-shot learning under support-query distribution shifts by generating hard examples and smoothing optimal transportation plans.

Contribution

The paper proposes PGADA, a new adversarial augmentation technique combined with regularized optimal transportation to address support-query shift in few-shot learning.

Findings

Significant performance improvements over state-of-the-art methods.

Effective in handling distribution shifts between support and query sets.

Robust across multiple benchmark datasets.

Abstract

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.