Can We Obtain Fairness For Free?

Can We Obtain Fairness For Free?
复制标题

我们能免费获得公平吗?

DOI:
10.1145/3461702.3462614
复制
发表时间:
2021
期刊:
and Society 2021
影响因子:
--
通讯作者:
Foulds, James R.
Foulds, James R.
中科院分区:
--
文献类型:
--
作者:
Islam, Rashidul;Pan, Shimei;Foulds, James R.

文献摘要

参考文献

被引文献

相似文献

越来越多的人意识到,人工智能和机器学习系统在某些情况下可以学会以不公平和歧视性的方式行事,并产生有害的后果。然而,尽管进行了大量的研究,但确保AI公平性的技术尚未在真实的系统中广泛部署。其中一个主要障碍是传统的观点,即公平性会带来预测性性能指标的成本,例如准确性,这可能会影响组织的底线。在本文中,我们将仔细研究这一问题。显然,公平/绩效权衡是存在的,但它们是不可避免的吗?与传统观点相反,我们发现,在不牺牲预测性能的情况下,提高训练模型的公平性通常是可能的,实际上是直接的。我们系统地研究了公平学习算法在一系列基准数据集上的行为,表明通过合理的超参数选择策略,可以在一定程度上提高公平性,而不会损失(甚至提高)预测性能。我们的研究结果揭示了增加公平人工智能方法部署的途径,具有潜在的积极现实影响。
There is growing awareness that AI and machine learning systems can in some cases learn to behave in unfair and discriminatory ways with harmful consequences. However, despite an enormous amount of research, techniques for ensuring AI fairness have yet to see widespread deployment in real systems. One of the main barriers is the conventional wisdom that fairness brings a cost in predictive performance metrics such as accuracy which could affect an organization's bottom-line. In this paper we take a closer look at this concern. Clearly fairness/performance trade-offs exist, but are they inevitable? In contrast to the conventional wisdom, we find that it is frequently possible, indeed straightforward, to improve on a trained model's fairness without sacrificing predictive performance. We systematically study the behavior of fair learning algorithms on a range of benchmark datasets, showing that it is possible to improve fairness to some degree with no loss (or even an improvement) in predictive performance via a sensible hyper-parameter selection strategy. Our results reveal a pathway toward increasing the deployment of fair AI methods, with potentially substantial positive real-world impacts.
DOI: --
发表时间: 2016-11
期刊: --
影响因子: --
作者:
Gilles Louppe;M. Kagan;Kyle Cranmer
通讯作者: Gilles Louppe;M. Kagan;Kyle Cranmer
DOI: --
发表时间: 2020
期刊: International Conference on Machine Learning
影响因子: --
作者:
Sanghamitra Dutta;Dennis Wei;Hazar Yueksel;Pin;Sijia Liu;Kush R. Varshney
通讯作者: Kush R. Varshney
对各种测试偏差定义的统计和伦理影响的批判性分析。
DOI: --
发表时间: 1976
期刊:
影响因子: --
作者:
J. Hunter;F. Schmidt
通讯作者: F. Schmidt
具有公平性约束的贝叶斯优化
DOI: --
发表时间: 2020
期刊:
影响因子: --
作者:
Valerio Perrone;Michele Donini
通讯作者: Michele Donini
DOI: --
发表时间: 2016
期刊:
影响因子: --
作者:
K. Crawford
通讯作者: K. Crawford