[论文解读] Exploring the Trade-off between Plausibility, Change Intensity and Adversarial Power in Counterfactual Explanations using Multi-objective Optimization

Javier Del Ser, Alejandro Barredo-Arrieta+3 more|arXiv (Cornell University)|May 20, 2022

Adversarial Robustness in Machine Learning被引用 4

一句话总结

本文提出了一种多目标优化框架，用于生成平衡真实性、变化强度和对抗性能力的反事实解释，采用基于生成对抗网络（GAN）的数据分布模型和多目标求解器。该框架在图像和3D数据上生成了更具可信度、直观性和可操作性的反事实解释，支持非专家用户检测偏差并提升模型可解释性。

ABSTRACT

There is a broad consensus on the importance of deep learning models in tasks involving complex data. Often, an adequate understanding of these models is required when focusing on the transparency of decisions in human-critical applications. Besides other explainability techniques, trustworthiness can be achieved by using counterfactuals, like the way a human becomes familiar with an unknown process: by understanding the hypothetical circumstances under which the output changes. In this work we argue that automated counterfactual generation should regard several aspects of the produced adversarial instances, not only their adversarial capability. To this end, we present a novel framework for the generation of counterfactual examples which formulates its goal as a multi-objective optimization problem balancing three different objectives: 1) plausibility, i.e., the likeliness of the counterfactual of being possible as per the distribution of the input data; 2) intensity of the changes to the original input; and 3) adversarial power, namely, the variability of the model's output induced by the counterfactual. The framework departs from a target model to be audited and uses a Generative Adversarial Network to model the distribution of input data, together with a multi-objective solver for the discovery of counterfactuals balancing among these objectives. The utility of the framework is showcased over six classification tasks comprising image and three-dimensional data. The experiments verify that the framework unveils counterfactuals that comply with intuition, increasing the trustworthiness of the user, and leading to further insights, such as the detection of bias and data misrepresentation.

研究动机与目标

• 通过生成直观且可操作的反事实解释，填补非专家用户在模型可解释性方面的空白。
• 将反事实生成形式化为一个多目标优化问题，以平衡真实性、变化强度和对抗性能力。
• 通过生成反映真实世界数据分布并揭示潜在偏差的反事实解释，提升模型的可信度。
• 为模型行为提供更深入的洞察，包括数据误代表现象以及对对抗性扰动的脆弱性。

提出的方法

• 训练一个生成对抗网络（GAN）以建模输入数据分布，从而生成具有真实感的反事实解释。
• 通过在条件向量上对GAN的判别器进行条件化，引导生成具有期望输出变化的反事实解释。
• 采用多目标优化求解器，寻找在三个目标之间实现平衡的反事实解释：真实性、变化强度和对抗性能力。
• 通过GAN生成与原始数据分布接近的逼真样本，评估真实性。
• 变化强度通过原始输入与反事实解释之间的L2距离进行度量。
• 对抗性能力通过将反事实解释输入目标模型后输出变化的幅度进行量化。

实验结果

研究问题

• RQ1 反事实生成是否本质上是一个涉及冲突目标的多目标优化问题？
• RQ2 生成的反事实解释是否符合基于数据和任务背景的直观预期？
• RQ3 多准则反事实解释能否揭示训练集中隐藏的偏差或数据误代表现象？
• RQ4 真实性、变化强度和对抗性能力之间的权衡如何影响用户信任度与可解释性？

主要发现

• 六个实验中帕累托前沿的近似结果证实，反事实生成受多个常存在冲突的目标支配，必须通过多目标优化求解。
• 生成的反事实解释表现出真实可信的变化——如颜色变化或结构强调——与人类直觉和任务特定语义一致。
• 该框架成功揭示了组合性偏差和数据误代表现象，特别是在属性-类别关系方面，表明潜在的脆弱性。
• 发现对抗性能力高且变化强度低的反事实解释更具可操作性，同时在所有数据集上均保持了高真实性。
• 该框架使非专家用户能够通过提供直观且可操作的模型预测替代方案，更好地理解模型行为。