Wasserstein Fair Classification

Jiang, Ray; Pacchiano, Aldo; Stepleton, Tom; Jiang, Heinrich; Chiappa, Silvia

doi:10.48550/arxiv.1907.12059

Cited by 8 publications

(8 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The above definition naturally extends to the multi-class setting the DP considered in binary classification [2,6,13,16,20] . Intuitively, when fairness is required, two important aspects of a classifier need to be controlled: the misclassification risk R(•) and the unfairness, evaluated as follows.…”

Section: Multi-class Classification With Demographic Paritymentioning

confidence: 99%

Fairness guarantee in multi-class classification

Denis,

Elie,

Hebiri

et al. 2021

Preprint

View full text Add to dashboard Cite

Algorithmic Fairness is an established area of machine learning, willing to reduce the influence of biases in the data. Yet, despite its wide range of applications, very few works consider the multi-class classification setting from the fairness perspective. We address this question by extending the definition of Demographic Parity to the multi-class problem while specifying the corresponding expression of the optimal fair classifier. This suggests a plug-in data-driven procedure, for which we establish theoretical guarantees. Specifically, we show that the enhanced estimator mimics the behavior of the optimal rule, both in terms of fairness and risk. Notably, fairness guarantee is distribution-free. We illustrate numerically the quality of our algorithm. The procedure reveals to be much more suitable than an alternative approach enforcing fairness constraints on the score associated to each class. This shows that our method is empirically very effective in fair decision making on both synthetic and real datasets.

show abstract

Section: Multi-class Classification With Demographic Paritymentioning

confidence: 99%

Fairness guarantee in multi-class classification

Denis,

Elie,

Hebiri

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Madras et al [2018] proposed a generalized framework to learn adversarially fair and transferable representations and suggests using the label information in the adversary to learn equalized odds or equal opportunity representations in the classification setting. Apart from adversarial representation, recent work also proposed to use distance metrics, e.g., the maximum mean discrepancy [Louizos et al, 2015] and the Wasserstein distance [Jiang et al, 2019] to remove group-related information. On the theoretical side, existing works have also studied the tradeoff between accuracy and statistical parity [McNamara et al, 2019, Dutta et al, 2020, but the majority of the existing works focus on the classification setting, with notable exceptions of [Le Gouic et al, 2020, Zhao et al, 2020b.…”

Section: Related Workmentioning

confidence: 99%

“…Despite the ample needs of accuracy parity, most prior work limits its scope to studying the problem in binary classification settings [Hardt et al, 2016, Zafar et al, 2017b, Jiang et al, 2019. Compared to the accuracy disparity problem in classification settings, accuracy disparity in regression is a more challenging but less studied problem, due to the fact that many existing algorithmic techniques designed for classification cannot be extended in a straightforward way when the target variable is continuous .…”

Section: Introductionmentioning

confidence: 99%

Understanding and Mitigating Accuracy Disparity in Regression

Chi,

Tian,

Gordon

et al. 2021

Preprint

View full text Add to dashboard Cite

With the widespread deployment of large-scale prediction systems in high-stakes domains, e.g., face recognition, criminal justice, etc., disparity on prediction accuracy between different demographic subgroups has called for fundamental understanding on the source of such disparity and algorithmic intervention to mitigate it. In this paper, we study the accuracy disparity problem in regression. To begin with, we first propose an error decomposition theorem, which decomposes the accuracy disparity into the distance between marginal label distributions and the distance between conditional representations, to help explain why such accuracy disparity appears in practice. Motivated by this error decomposition and the general idea of distribution alignment with statistical distances, we then propose an algorithm to reduce this disparity, and analyze its game-theoretic optima of the proposed objective functions. To corroborate our theoretical findings, we also conduct experiments on five benchmark datasets. The experimental results suggest that our proposed algorithms can effectively mitigate accuracy disparity while maintaining the predictive power of the regression models. 1

show abstract

“…The goal of the adversary is to infer the group attribute as much as possible while the goal of the data owner is to remove information related to the group attribute and simultaneously to preserve utility-related information for accurate prediction. Apart from using adversarial classifiers to enforce group fairness, other distance metrics have also been used to learn fair representations, e.g., the maximum mean discrepancy (Louizos et al, 2015), and the Wasserstein-1 distance (Jiang et al, 2019). In contrast to these methods, in this paper we advocate for optimizing BER on both the target loss and adversary loss in order to simultaneously achieve accuracy parity and equalized odds.…”

Section: Related Workmentioning

confidence: 99%

Conditional Learning of Fair Representations

Han¹,

Coston²,

Adel³

et al. 2019

Preprint

View full text Add to dashboard Cite

We propose a novel algorithm for learning fair representations that can simultaneously mitigate two notions of disparity among different demographic subgroups. Two key components underpinning the design of our algorithm are balanced error rate and conditional alignment of representations. We show how these two components contribute to ensuring accuracy parity and equalized false-positive and false-negative rates across groups without impacting demographic parity. Furthermore, we also demonstrate both in theory and on two real-world experiments that the proposed algorithm leads to a better utility-fairness trade-off on balanced datasets compared with existing algorithms on learning fair representations.

show abstract

Wasserstein Fair Classification

Cited by 8 publications

References 19 publications

Fairness guarantee in multi-class classification

Fairness guarantee in multi-class classification

Understanding and Mitigating Accuracy Disparity in Regression

Conditional Learning of Fair Representations

Contact Info

Product

Resources

About