Adversarially Robust Few-Shot Learning: A Meta-Learning Approach
This addresses the vulnerability of few-shot learning methods to adversarial attacks, which is a critical issue for deploying AI in security-sensitive domains, though it appears incremental as it builds on existing meta-learning frameworks.
The paper tackled the problem of making few-shot learning models robust to adversarial examples, achieving far superior robust performance on few-shot image classification tasks like Mini-ImageNet and CIFAR-FS compared to robust transfer learning.
Previous work on adversarially robust neural networks for image classification requires large training sets and computationally expensive training procedures. On the other hand, few-shot learning methods are highly vulnerable to adversarial examples. The goal of our work is to produce networks which both perform well at few-shot classification tasks and are simultaneously robust to adversarial examples. We develop an algorithm, called Adversarial Querying (AQ), for producing adversarially robust meta-learners, and we thoroughly investigate the causes for adversarial vulnerability. Moreover, our method achieves far superior robust performance on few-shot image classification tasks, such as Mini-ImageNet and CIFAR-FS, than robust transfer learning.