On fair selection in the presence of implicit and differential variance

Emelianov, Vitalii; Gast, Nicolas; Gummadi, Krishna P.; Loiseau, Patrick

doi:10.1016/j.artint.2021.103609

Cited by 13 publications

(34 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Further, we show that this result also holds for generalizations of the implicit bias model where, for instance, the implicit bias experienced by individuals in multiple groups is different than the product of the implicit bias parameters of the groups they belong to (Theorem 3.4). Thus, these results imply that, unlike the setting of a single affected group studied in [57,20,36], non-intersectional constraints may be insufficient to completely mitigate the effects of implicit bias in the presence of intersections.…”

Section: Our Contributionsmentioning

confidence: 85%

“…Models of implicit bias and decision-making in the presence of implicit bias. A growing literature is studying decision making in the presence of implicit bias, ranging from works on set selection [57,36,38], ranking [20], to classification [12]. Among these, works on the set selection and ranking problems are directly related to our work.…”

Section: Related Workmentioning

confidence: 99%

“…While we also consider the model of implicit bias [20] introduced, the intersectional constraints we propose are different and do not depend on the, unknown, latent utilities. [36] study selection under a different model of bias, where the decision maker's observed utility has higher than average noise for individuals in the affected group. They consider a family of constraints and show that, for a single affected group, these constraints increase the latent utility.…”

Section: Related Workmentioning

confidence: 99%

“…Here, we consider lower-bound constraints, such as the Rooney Rule, which require the decision maker to select at least a specified number of candidates from each group. Recent works [57,20,36] on decision-making in the presence of implicit bias have shown that in the setting with a single group, such constraints can serve an additional purpose: to improve the latent utility of the subset selected by the decision maker. Motivated by this, we study the efficacy these constraints to improve the latent utility in the setting with multiple and intersectional groups, as discussed in Section 2.2.…”

Section: Selection Under Implicit Bias: Intersectional and Non-inters...mentioning

confidence: 99%

“…Recently, some works have analyzed the effectiveness of Rooney Rule type constraints for selection and ranking processes [57,20,36]. In particular, [57] study the effectiveness of the Rooney Rule for selection in the presence of implicit bias when there is a single affected group.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Selection in the Presence of Implicit Bias: The Advantage of Intersectional Constraints

Mehrotra¹,

Pradelski²,

Vishnoi³

2022

Preprint

View full text Add to dashboard Cite

In selection processes such as hiring, promotion, and college admissions, implicit bias toward sociallysalient attributes such as race, gender, or sexual orientation of candidates is known to produce persistent inequality and reduce aggregate utility for the decision maker. Interventions such as the Rooney Rule and its generalizations, which require the decision maker to select at least a specified number of individuals from each affected group, have been proposed to mitigate the adverse effects of implicit bias in selection. Recent works have established that such lower-bound constraints can be very effective in improving aggregate utility in the case when each individual belongs to at most one affected group. However, in several settings, individuals may belong to multiple affected groups and, consequently, face more extreme implicit bias due to this intersectionality. We consider independently drawn utilities and show that, in the intersectional case, the aforementioned non-intersectional constraints can only recover part of the total utility achievable in the absence of implicit bias. On the other hand, we show that if one includes appropriate lower-bound constraints on the intersections, almost all the utility achievable in the absence of implicit bias can be recovered. Thus, intersectional constraints can offer a significant advantage over a reductionist dimension-by-dimension non-intersectional approach to reducing inequality.

show abstract

Section: Our Contributionsmentioning

confidence: 85%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Selection Under Implicit Bias: Intersectional and Non-inters...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Selection in the Presence of Implicit Bias: The Advantage of Intersectional Constraints

Mehrotra¹,

Pradelski²,

Vishnoi³

2022

Preprint

View full text Add to dashboard Cite

show abstract

FAC-fed: Federated adaptation for fairness and concept drift aware stream classification

2023

View full text Add to dashboard Cite

Federated learning is an emerging collaborative learning paradigm of Machine learning involving distributed and heterogeneous clients. Enormous collections of continuously arriving heterogeneous data residing on distributed clients require federated adaptation of efficient mining algorithms to enable fair and high-quality predictions with privacy guarantees and minimal response delay. In this context, we propose a federated adaptation that mitigates discrimination embedded in the streaming data while handling concept drifts (FAC-Fed). We present a novel adaptive data augmentation method that mitigates client-side discrimination embedded in the data during optimization, resulting in an optimized and fair centralized server. Extensive experiments on a set of publicly available streaming and static datasets confirm the effectiveness of the proposed method. To the best of our knowledge, this work is the first attempt towards fairness-aware federated adaptation for stream classification, therefore, to prove the superiority of our proposed method over state-of-the-art, we compare the centralized version of our proposed method with three centralized stream classification baseline models (FABBOO, FAHT, CSMOTE). The experimental results show that our method outperforms the current methods in terms of both discrimination mitigation and predictive performance.

show abstract