Adversarial Robustness is at Odds with Lazy Training

Authors: Yunjuan Wang, Enayat Ullah, Poorya Mianjy, Raman Arora
Published: 2022-06-18 | Updated: 2022-10-17

Source: https://arxiv.org/abs/2207.00411

PDF: https://arxiv.org/pdf/2207.00411

AIにより推定されたラベル

アルゴリズム設計 敵対的サンプル 敵対的攻撃検出

※ こちらのラベルはAIによって自動的に追加されました。そのため、正確でないことがあります。
詳細は文献データベースについてをご覧ください。

Abstract

Recent works show that adversarial examples exist for random neural networks [Daniely and Schacham, 2020] and that these examples can be found using a single step of gradient ascent [Bubeck et al., 2021]. In this work, we extend this line of work to “lazy training” of neural networks – a dominant model in deep learning theory in which neural networks are provably efficiently learnable. We show that over-parametrized neural networks that are guaranteed to generalize well and enjoy strong computational guarantees remain vulnerable to attacks generated using a single step of gradient ascent.