FairPipes: Data Mutation Pipelines for Machine Learning Fairness - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2024

FairPipes: Data Mutation Pipelines for Machine Learning Fairness

Paul Temple
Gilles Perrouin

Résumé

Machine Learning (ML) models are ubiquitous in decisionmaking applications impacting citizens' lives: credit attribution, crime recidivism, etc. In addition to seeking high performance and generalization abilities, ensuring that ML models do not discriminate against citizens regarding their age, gender, or race is essential. To this end, researchers developed various fairness assessment techniques, comprising fairness metrics and mitigation approaches, notably at the model level. However, the sensitivity of ML models to fairness data perturbations has been less explored. This paper presents mutation-based pipelines to emulate fairness variations in the data once the model is deployed. FairPipes implements mutation operators that shuffle sensitive attributes, add new values, or affect their distribution. We evaluated FairPipes on seven ML models over three datasets. Our results highlight different fairness sensitivity behaviors across models, from the most sensitive perceptrons to the insensitive support vector machines. We also consider the role of model optimization in fairness performance, being variable across models. FairPipes automates fairness testing at deployment time, informing researchers and practitioners on the fairness sensitivity evolution of their ML models. CCS CONCEPTS • Software and its engineering → Software testing and debugging; • Computing methodologies → Machine learning.
Fichier principal
Vignette du fichier
AST_2024___Fairness_Assessment_via_Mutation_Testing.pdf (1.28 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04440201 , version 1 (05-02-2024)

Licence

Paternité

Identifiants

Citer

Camille Molinier, Paul Temple, Gilles Perrouin. FairPipes: Data Mutation Pipelines for Machine Learning Fairness. AST 2024 - 5th ACM/IEEE International Conference on Automation of Software Test, Apr 2024, Lisbonne, Portugal. pp.1-11, ⟨10.1145/3644032.3644465⟩. ⟨hal-04440201⟩
100 Consultations
75 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More