Using Copies to Remove Sensitive Data: A Case Study on Fair Superhero Alignment Prediction

Producció científica: Capítol de llibreContribució a congrés/conferènciaAvaluat per experts

4 Cites (Scopus)

Resum

Ensuring classification models are fair with respect to sensitive data attributes is a crucial task when applying machine learning models to real-world problems. Particularly in company production environments, where the decision output by models may have a direct impact on individuals and predictive performance should be maintained over time. In this article, build upon [17], we propose copies as a technique to mitigate the bias of trained algorithms in circumstances where the original data is not accessible and/or the models cannot be re-trained. In particular, we explore a simple methodology to build copies that replicate the learned decision behavior in the absence of sensitive attributes. We validate this methodology in the low-sensitive problem of superhero alignment. We demonstrate that this naïve approach to bias reduction is feasible in this problem and argue that copies can be further exploited to embed models with desiderata such as fair learning.

Idioma originalAnglès
Títol de la publicacióPattern Recognition and Image Analysis - 9th Iberian Conference, IbPRIA 2019, Proceedings
EditorsAythami Morales, Julian Fierrez, José Salvador Sánchez, Bernardete Ribeiro
EditorSpringer
Pàgines182-193
Nombre de pàgines12
ISBN (imprès)9783030313319
DOIs
Estat de la publicacióPublicada - 2019
Publicat externament
Esdeveniment9th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2019 - Madrid, Spain
Durada: 1 de jul. 20194 de jul. 2019

Sèrie de publicacions

NomLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volum11867 LNCS
ISSN (imprès)0302-9743
ISSN (electrònic)1611-3349

Conferència

Conferència9th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2019
País/TerritoriSpain
CiutatMadrid
Període1/07/194/07/19

Fingerprint

Navegar pels temes de recerca de 'Using Copies to Remove Sensitive Data: A Case Study on Fair Superhero Alignment Prediction'. Junts formen un fingerprint únic.

Com citar-ho