RT info:eu-repo/semantics/article
T1 Attraction-repulsion clustering: a way of promoting diversity linked to demographic parity in fair clustering
A1 Barrio Tellado, Eustasio del
A1 Inouzhe Valdés, Hristo
A1 Loubes, Jean-Michel
K1 Diversity enhancing clustering
K1 Demographic parity
K1 Fair clustering
K1 Hierarchical clustering
K1 Kernel methods
K1 Multidimensional scaling
K1 12 Matemáticas
AB We consider the problem of diversity enhancing clustering, i.e, developing clusteringmethods which produce clusters that favour diversity with respect to a set of pro-tected attributes such as race, sex, age, etc. In the context of fair clustering, diversityplays a major role when fairness is understood as demographic parity. To promotediversity, we introduce perturbations to the distance in the unprotected attributes thataccount for protected attributes in a way that resembles attraction-repulsion of chargedparticles in Physics. These perturbations are defined through dissimilarities with atractable interpretation. Cluster analysis based on attraction-repulsion dissimilaritiespenalizes homogeneity of the clusters with respect to the protected attributes and leadsto an improvement in diversity. An advantage of our approach, which falls into a pre-processing set-up, is its compatibility with a wide variety of clustering methods andwhit non-Euclidean data. We illustrate the use of our procedures with both syntheticand real data and provide discussion about the relation between diversity, fairness, andcluster structure.
PB Springer
SN 1862-5347
YR 2022
FD 2022
LK https://uvadoc.uva.es/handle/10324/57040
UL https://uvadoc.uva.es/handle/10324/57040
LA eng
NO Advances in Data Analysis and Classification, 2022,
NO Producción Científica
DS UVaDOC
RD 18-nov-2024