一句话：提出基于特征重要性与扰动潜力的种子优先级排序方法 FIPSER，提升 DNN 公平性测试中个体歧视实例对挖掘的效果与效率。
深度神经网络日益融入人类社会，但其公平性问题引发关注。已有工作使用因果公平性度量衡量模型公平性，并通过搜索算法挖掘个体歧视实例对（IDIP），再用修正后的 IDIP 重训练模型来缓解偏见。然而，受效率限制，这类方法可用作种子的样本数量通常有限，且不同种子产生的 IDIP 数量差异显著，因此如何选择和排序种子尚未被充分研究。本文首先研究不同数据集与敏感属性下 IDIP 数量的不平衡，说明种子选择与排序的必要性。随后提出 FIPSER，一种基于特征重要性和扰动潜力的种子优先级排序方法。实验表明，将其应用于当前最先进的 IDIP 挖掘方法时，平均可提升效果 45%、效率 11%。
As a rapidly evolving AI technology, deep neural networks are becoming increasingly integrated into human society, yet raising concerns about fairness issues. Previous studies have proposed a metric called causal fairness to measure the fairness of machine learning models and proposed some search algorithms to mine individual discrimination instance pairs (IDIPs). Fairness issues can be alleviated by retraining models with corrected IDIPs. However, the number of samples that are used as seeds for these methods is often limited due to the pursuit of efficiency. In addition, the quantity of IDIPs generated on different seeds varies, so it makes sense to select appropriate samples as seeds, which has not been sufficiently considered in past studies. In this paper, we study the imbalance in IDIP quantities for various datasets and sensitive attributes, highlighting the need for selecting and ranking seed samples. Then, we proposed FIPSER, a feature importance and perturbation potential-based seed prioritization method. Our experimental results show that, on average, when applied to the current state-of-the-art method of IDIP mining, FIPSER can improve its effectiveness by 45% and efficiency by 11%.CCS CONCEPTS• Computing methodologies → Artificial intelligence; • Software and its engineering;
@inproceedings{ChenZZZ00P24,
title = {FIPSER: Improving Fairness Testing of DNN by Seed Prioritization},
author = {Junwei Chen and Yueling Zhang and Lingfeng Zhang and Min Zhang and Chengcheng Wan and Ting Su and Geguang Pu},
booktitle = {ASE},
year = {2024},
doi = {10.1145/3691620.3695486}
}