Attacking Few-Shot Classifiers with Adversarial Support Sets

Elre Talea Oldewage, John F Bronskill, Richard E Turner

Keywords: meta-learning, few-shot learning, adversarial attacks, poisoning

Abstract: Few-shot learning systems, especially those based on meta-learning, have recently made significant advances, and are now being considered for real world problems in healthcare, personalization, and science. In this paper, we examine the robustness of such deployed few-shot learning systems when they are fed an imperceptibly perturbed few-shot dataset, showing that the resulting predictions on test inputs can become worse than chance. This is achieved by developing a novel Adversarial Support Set Attack which crafts a poisoned set of examples. When even a small subset of malicious data points is inserted into the support set of a meta-learner, accuracy is significantly reduced. For example, the average classification accuracy of CNAPs on the Aircraft dataset in the META-DATASET benchmark drops from 69.2% to 9.1% when only 20% of the support set is poisoned by imperceptible perturbations. We evaluate the new attack on a variety of few-shot classification algorithms including MAML, prototypical networks, and CNAPs, on both small scale (miniImageNet) and large scale (META-DATASET) few-shot classification problems. Interestingly, adversarial support sets produced by attacking a meta-learning based few-shot classifier can also reduce the accuracy of a fine-tuning based few-shot classifier when both models use similar feature extractors.

One-sentence Summary: We introduce an effective, novel adversarial attack called an Adversarial Support Set Attack that poisons the support set of trained few-shot learners to cause failure with high probability on unseen queries at test time.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Reviewed Version (pdf): https://openreview.net/references/pdf?id=74yDxY5AmI