A General Framework for Fair Regression.

A General Framework for Fair Regression.
复制标题

DOI:
10.3390/e21080741
复制
发表时间:
2019-07-29
期刊:
Entropy (Basel, Switzerland)
影响因子:
--
通讯作者:
Roberts S
Roberts S
中科院分区:
其他
文献类型:
--
作者:
Fitzsimons J;Al Ali A;Osborne M;Roberts S

文献摘要

参考文献

被引文献

相似文献

公平,通过它的多种形式和定义,已经成为机器学习社区面临的一个重要问题。在这项工作中,我们考虑如何将群体公平约束融入到核回归方法中,适用于高斯过程、支持向量机、神经网络回归和决策树回归。此外,我们重点考察了将这些约束纳入决策树回归的效果,并将其直接应用于随机森林和增强树以及其他广泛流行的推理技术。我们证明了对于这样的模型,存储和计算的复杂性的顺序是保持的,并且根据树叶的数量将预期的扰动与模型紧密地绑定在一起。重要的是,该方法适用于训练模型,因此可以很容易地应用于当前使用的模型,并且只需要在训练数据上使用分组标签。
Fairness, through its many forms and definitions, has become an important issue facing the machine learning community. In this work, we consider how to incorporate group fairness constraints into kernel regression methods, applicable to Gaussian processes, support vector machines, neural network regression and decision tree regression. Further, we focus on examining the effect of incorporating these constraints in decision tree regression, with direct applications to random forests and boosted trees amongst other widespread popular inference techniques. We show that the order of complexity of memory and computation is preserved for such models and tightly binds the expected perturbations to the model in terms of the number of leaves of the trees. Importantly, the approach works on trained models and hence can be easily applied to models in current use and group labels are only required on training data.
DOI: 10.1007/s10994-006-6226-1
发表时间: 2006-04-01
期刊: MACHINE LEARNING
影响因子: 7.5
作者:
Geurts, P;Ernst, D;Wehenkel, L
通讯作者: Wehenkel, L
DOI: 10.1111/j.1365-2656.2008.01390.x
发表时间: 2008-07-01
影响因子: 4.8
作者:
Elith, J.;Leathwick, J. R.;Hastie, T.
通讯作者: Hastie, T.
预测累犯的准确性,公平性和局限性。
DOI: 10.1126/sciadv.aao5580
发表时间: 2018-01
期刊: Science advances
影响因子: 13.6
作者:
Dressel J;Farid H
通讯作者: Farid H
DOI: 10.1007/s10618-010-0190-x
发表时间: 2010-09-01
影响因子: 4.8
作者:
Calders, Toon;Verwer, Sicco
通讯作者: Verwer, Sicco