哪些模型具有感知对齐梯度？一种通过流形外鲁棒性的解释
Which Models have Perceptually-Aligned Gradients? An Explanation via Off-Manifold Robustness
AI总结：
该研究针对鲁棒视觉模型的感知对齐梯度机制不明的问题，通过流形外鲁棒性首次给出理论解释，验证了多种鲁棒训练模型的符合度，并划分出三种影响感知对齐与准确率的鲁棒性状态。
摘要：
鲁棒计算机视觉模型的显著特性之一是其输入梯度通常与人类感知对齐，文献中称之为感知对齐梯度（perceptually-aligned gradients, PAGs）。尽管仅针对分类任务训练，PAGs使鲁棒模型具备了基础的生成能力，包括图像生成、去噪和图像修复。然而，这些现象背后的潜在机制仍不明确。在本研究中，我们通过**流形外鲁棒性（off-manifold robustness）**首次对PAGs给出了解释，该特性指模型在数据流形外的鲁棒性必须高于其在流形上的鲁棒性。我们首先从理论上证明，流形外鲁棒性会使输入梯度近似位于数据流形上，从而解释了其感知对齐特性。随后我们证明贝叶斯最优模型满足流形外鲁棒性，并通过实验证实，通过梯度范数正则化、随机平滑（randomized smoothing）以及基于投影梯度下降的对抗训练得到的鲁棒模型也满足该特性。我们通过计算模型梯度与生成模型梯度的相似性来量化梯度的感知对齐程度，结果表明流形外鲁棒性与感知对齐程度具有良好的相关性。最后，基于流形上和流形外的鲁棒性水平，我们确定了三种会同时影响感知对齐和模型准确率的鲁棒性状态：弱鲁棒性、贝叶斯对齐鲁棒性和过度鲁棒性。代码可在https://github.com/tml-tuebingen/pags获取。
英文摘要：
One of the remarkable properties of robust computer vision models is that their input-gradients are often aligned with human perception, referred to in the literature as perceptually-aligned gradients (PAGs). Despite only being trained for classification, PAGs cause robust models to have rudimentary generative capabilities, including image generation, denoising, and in-painting. However, the underlying mechanisms behind these phenomena remain unknown. In this work, we provide a first explanation of PAGs via \emph{off-manifold robustness}, which states that models must be more robust off- the data manifold than they are on-manifold. We first demonstrate theoretically that off-manifold robustness leads input gradients to lie approximately on the data manifold, explaining their perceptual alignment. We then show that Bayes optimal models satisfy off-manifold robustness, and confirm the same empirically for robust models trained via gradient norm regularization, randomized smoothing, and adversarial training with projected gradient descent. Quantifying the perceptual alignment of model gradients via their similarity with the gradients of generative models, we show that off-manifold robustness correlates well with perceptual alignment. Finally, based on the levels of on- and off-manifold robustness, we identify three different regimes of robustness that affect both perceptual alignment and model accuracy: weak robustness, bayes-aligned robustness, and excessive robustness. Code is available at \url{https://github.com/tml-tuebingen/pags}.
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。
