Toward Unified Data and Algorithm Fairness via Adversarial Data Augmentation and Adaptive Model Fine-tuning
Toward Unified Data and Algorithm Fairness via Adversarial Data Augmentation and Adaptive Model Fine-tuning
复制标题
DOI：
10.1109/icdm54844.2022.00174
复制
发表时间：
2022-11
期刊：
影响因子：
--
通讯作者：
Yanfu Zhang;Runxue Bao;Jian Pei;Heng Huang
中科院分区：
文献类型：
--
作者：
Yanfu Zhang;Runxue Bao;Jian Pei;Heng Huang
最近对于有偏差数据的算法公平性存在一些研究兴趣。针对这个问题设计了各种各样的预处理、处理中以及后处理方法。然而，这些方法仅仅针对数据不公平和算法不公平。在本文中，我们提出一种新的处理中方法来拓宽公平性方法的应用场景，它能够同时处理这两种偏差来源。由于现代深度模型因庞大的训练数据和复杂的结构而从头开始训练成本很高，我们提出一个增强和微调框架。首先，我们设计一种对抗性攻击来生成与受保护属性分离的加权样本。接下来，我们在有偏差的模型中识别公平的子结构，并通过权重重新激活来微调模型。最后，我们为增强和微调提供一个可选的联合训练方案。我们的方法可以与多种公平性度量相结合。我们对我们的方法和一些相关的基线进行基准测试，以展示其优势和可扩展性。在几个标准数据集上的实验结果表明，我们的方法能够有效地学习公平增强，并取得优于最先进基线的结果。我们的方法对不同类型的数据也有很好的泛化能力。
There is some recent research interest in algorithmic fairness for biased data. There are a variety of pre-, in-, and post-processing methods designed for this problem. However, these methods are exclusively targeting data unfairness and algorithmic unfairness. In this paper, we propose a novel intra-processing method to broaden the application scenario of fairness methods, which can simultaneously address the two bias sources. Since training modern deep models from scratch is expensive due to the enormous training data and the complicated structures, we propose an augmentation and fine-tuning framework. First, we design an adversarial attack to generate weighted samples disentangled with the protected attribute. Next, we identify the fair sub-structure in the biased model and fine-tune the model via weight reactivation. At last, we provide an optional joint training scheme for the augmentation and the fine-tuning. Our method can be combined with a variety of fairness measures. We benchmark our method and some related baselines to show the advantage and the scalability. Experimental results on several standard datasets demonstrate that our approach can effectively learn fair augmentation and achieve superior results to the state-of-the-art baselines. Our method also generalizes well to different types of data.