arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2408.09672cs.LGmath.OCstat.ML
收藏

对抗鲁棒学习的正则化

Regularization for Adversarial Robust Learning

Georgia Institute of Technology(佐治亚理工学院);The University of Texas at Austin(得克萨斯大学奥斯汀分校)

Jie Wang, Rui Gao, Yao Xie

AI总结:

针对基于$\infty$-Wasserstein度量的分布鲁棒风险计算不可行问题,提出一种整合$\phi$-散度正则化的对抗训练方法,利用带偏置预言机的随机梯度算法实现近最优样本复杂度,并在多种学习任务中达到最先进的防御性能。

摘要:

尽管人工神经网络在现实应用中日益普及,但其易受对抗攻击的脆弱性仍是一个重大隐患,这促使我们研究机器学习模型的鲁棒性。尽管各种启发式方法旨在使用 $\infty$-Wasserstein 度量优化分布鲁棒风险,但这种鲁棒性概念常面临计算不可行性。为解决计算挑战,我们开发了一种新颖的对抗训练方法,将 $\phi$-散度正则化整合到分布鲁棒风险函数中。与原始公式相比,这种正则化在计算上带来了显著改善。我们开发了带有偏置预言机的随机梯度方法来高效求解此问题,实现了近最优的样本复杂度。此外,我们通过考虑正则化参数和鲁棒性水平的各种缩放机制,确立了其正则化效应,并证明其与正则化经验风险最小化框架渐近等价。这些机制产生梯度范数正则化、方差正则化,或在两极之间进行插值的平滑梯度范数正则化。我们在监督学习、强化学习和上下文学习中数值验证了所提方法,并展示了其针对各种对抗攻击的最先进性能。

英文摘要:

Despite the growing prevalence of artificial neural networks in real-world applications, their vulnerability to adversarial attacks remains a significant concern, which motivates us to investigate the robustness of machine learning models. While various heuristics aim to optimize the distributionally robust risk using the $\infty$-Wasserstein metric, such a notion of robustness frequently encounters computation intractability. To tackle the computational challenge, we develop a novel approach to adversarial training that integrates $ϕ$-divergence regularization into the distributionally robust risk function. This regularization brings a notable improvement in computation compared with the original formulation. We develop stochastic gradient methods with biased oracles to solve this problem efficiently, achieving the near-optimal sample complexity. Moreover, we establish its regularization effects and demonstrate it is asymptotic equivalence to a regularized empirical risk minimization framework, by considering various scaling regimes of the regularization parameter and robustness level. These regimes yield gradient norm regularization, variance regularization, or a smoothed gradient norm regularization that interpolates between these extremes. We numerically validate our proposed method in supervised learning, reinforcement learning, and contextual learning and showcase its state-of-the-art performance against various adversarial attacks.

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。

↑