arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2202.04347cs.LG
收藏

梯度法可证明收敛于非鲁棒网络

Gradient Methods Provably Converge to Non-Robust Networks

TTI Chicago(芝加哥丰田技术研究所);Hebrew University of Jerusalem(耶路撒冷希伯来大学);Weizmann Institute of Science(魏茨曼科学研究所)

Gal Vardi, Gilad Yehudai, Ohad Shamir

摘要:

尽管进行了大量的研究,但仍然不清楚为什么神经网络如此容易受到对抗性样本的影响。在这项工作中,我们确定了自然环境,其中使用梯度流训练的深度$2$ ReLU网络可证明是非鲁棒的(易受小的对抗性$\ell_2$扰动的影响),即使正确分类训练数据集的鲁棒网络存在。也许令人惊讶的是,我们证明了众所周知的隐式偏差对利润最大化诱导非鲁棒网络的偏见,通过证明每个网络满足KKT条件的最大利润问题是非鲁棒的。

英文摘要:

Despite a great deal of research, it is still unclear why neural networks are so susceptible to adversarial examples. In this work, we identify natural settings where depth-$2$ ReLU networks trained with gradient flow are provably non-robust (susceptible to small adversarial $\ell_2$-perturbations), even when robust networks that classify the training dataset correctly exist. Perhaps surprisingly, we show that the well-known implicit bias towards margin maximization induces bias towards non-robust networks, by proving that every network which satisfies the KKT conditions of the max-margin problem is non-robust.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑