[论文解读] Does Adversarial Transferability Indicate Knowledge Transferability

Kaizhao Liang, Jacky Y. Zhang+2 more|arXiv (Cornell University)|May 4, 2021

Adversarial Robustness in Machine Learning参考文献 44被引用 7

一句话总结

本文研究了对抗性迁移性是否在深度神经网络中指示知识迁移性，提出了一套理论框架，识别出充分条件（尤其是仿射组合），以证明高对抗性迁移性可推导出高知识迁移性。在多种数据集上的实证结果表明二者存在强烈正相关，理论在实践中得到验证。

ABSTRACT

Despite the immense success that deep neural networks (DNNs) have achieved, \emph{adversarial examples}, which are perturbed inputs that aim to mislead DNNs to make mistakes, have recently led to great concerns. On the other hand, adversarial examples exhibit interesting phenomena, such as \emph{adversarial transferability}. DNNs also exhibit knowledge transfer, which is critical to improving learning efficiency and learning in domains that lack high-quality training data. To uncover the fundamental connections between these phenomena, we investigate and give an affirmative answer to the question: \emph{does adversarial transferability indicate knowledge transferability?} We theoretically analyze the relationship between adversarial transferability and knowledge transferability, and outline easily checkable sufficient conditions that identify when adversarial transferability indicates knowledge transferability. In particular, we show that composition with an affine function is sufficient to reduce the difference between two models when they possess high adversarial transferability. Furthermore, we provide empirical evaluation for different transfer learning scenarios on diverse datasets, showing a strong positive correlation between the adversarial transferability and knowledge transferability, thus illustrating that our theoretical insights are predictive of practice.

研究动机与目标

• 探究深度神经网络中对抗性迁移性与知识迁移性之间基本关系。
• 确定高对抗性迁移性是否可作为高知识迁移性的指示器。
• 推导理论上成立且易于检验的充分条件，以确保对抗性迁移性蕴含知识迁移性。
• 在多种迁移学习场景和数据集上实证验证理论洞见。

提出的方法

• 使用数学建模对对抗性迁移性与知识迁移性之间的关系进行理论分析。
• 识别出仿射变换组合为在对抗性迁移性较高时减少模型差异的充分条件。
• 推导出高对抗性迁移性蕴含高知识迁移性的条件。
• 在多样化的数据集和迁移学习设置中进行实证评估，以检验对抗性迁移性与知识迁移性之间的相关性。
• 使用标准深度学习基准和迁移学习协议，评估在不同模型与数据配置下的性能。

实验结果

研究问题

• RQ1 两模型之间高对抗性迁移性是否表明高知识迁移性？
• RQ2 确保对抗性迁移性蕴含知识迁移性的理论条件是什么？
• RQ3 仿射变换组合是否能减少具有高对抗性迁移性的模型之间的差异？
• RQ4 在不同数据集和迁移场景中，对抗性迁移性与知识迁移性之间的实证相关性有多强？

主要发现

• 在多样化数据集和迁移学习场景中，对抗性迁移性与知识迁移性之间存在强烈正相关。
• 与仿射函数组合是当模型表现出高对抗性迁移性时减少模型差异的充分条件。
• 论文推导出的理论条件经实证评估验证，具有对现实世界性能的预测能力。
• 在所识别的充分条件下，高对抗性迁移性可可靠地指示高知识迁移性。
• 研究结果表明，对抗性迁移性可作为评估模型设计与训练中知识迁移性的实用代理指标。