arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.20971cs.LGcs.AI
收藏

顺序重要吗:连接鲁棒性定律与鲁棒泛化

Does Order Matter : Connecting The Law of Robustness to Robust Generalization

Ashoka University(阿什oka大学);Truth Audit Labs(真相审计实验室);Indian Statistical Institute(印度统计研究所)

Mihir More, Aritra Das, Jaee Ponde, Himadri Mandal, Vishnu Varadarajan, Debayan Gupta

AI总结:

本文通过全局和局部Rademacher复杂度,将鲁棒性定律(Lipschitz常数下界)与鲁棒泛化误差联系起来,证明了对任意数据分布,全局Lipschitz界阶不变,而局部Lipschitz界阶随扰动半径和局部浓度项变化。

摘要:

Bubeck和Selke(2021)将鲁棒性定律与鲁棒泛化误差之间的联系作为一个开放问题提出。鲁棒性定律指出,过参数化对于模型实现鲁棒插值是必要的,即插值函数必须是Lipschitz的。Wu等人(2023)将该定律推广到任意数据分布,证明Lipschitz常数满足$L = Ω(n^{1/d})$。另一方面,鲁棒泛化研究小的鲁棒训练损失是否意味着小的鲁棒测试损失。这可以使用统计学习技术(如Rademacher复杂度)来研究,其中鲁棒损失类的Rademacher复杂度的界意味着函数类Lipschitz性的界。我们利用这一联系,明确地将两者联系起来,适用于任意数据分布。(i) 我们证明,在考虑鲁棒损失类的全局Rademacher复杂度时,Lipschitz界的阶保持不变。(ii) 在局部尺度上,即对于具有小经验误差的函数子集,Lipschitz界的阶随扰动半径$ρ$和局部浓度项$\sqrt{r/n}$变化。

英文摘要:

Bubeck and Selke (2021) propose the connection between the Law of Robustness and robust generalization error as an open problem. The Law of Robustness states that overparameterization is necessary for models to interpolate robustly, i.e., the interpolating function is required to be Lipschitz. Wu et al. (2023) extend this law to arbitrary data distributions, proving that the Lipschitz constant satisfies $L = Ω(n^{1/d})$. Robust generalization, on the other hand, asks whether small robust training loss implies small robust test loss. This can be studied using statistical learning techniques such as Rademacher complexities, where a bound on the Rademacher complexity of the robust loss class implies a bound on the Lipschitzness of the function class. We use this connection to explicitly link the two for arbitrary data distributions. (i) We prove that the order of the Lipschitz bound remains the same when considering the global Rademacher complexity of robust loss classes. (ii) At the local scale, i.e., for subsets of functions with small empirical error, the order of the Lipschitz bound changes with the perturbation radius $ρ$ and the localized concentration term $\sqrt{r/n}$.

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。

↑