Hierarchical clustering for discrimination discovery: a top-down approach

Neda Nasiriani, Anna Squicciarini, Zara Saldanha, Sanchit Goel, Nicola Zannone

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

Today, data is an essential part of many decision-making processes in businesses and social life through the use of various machine learning techniques. These methods can easily perpetuate human bias in the data and result in discrimination. Despite a growing interest in data discrimination discovery and removal, to date there is a lack of a general and robust framework to distinguish discriminatory decision-making processes from non-discriminatory ones. In this work, we present a generic framework that helps detect possible discrimination by analyzing historical data and associated decisions using a top-down unsupervised approach, which we refer to as hierarchical clustering. Our approach is highly adaptive as it gradually 'learns' users' inherent groups, and clusters their records using cohesiveness and density of points in the dataset. Moreover, we propose a progressive attribute-selection method to choose statistically relevant attributes, thus reducing the effect of noise. Finally, we adopt a recursive notion of cluster profile that is homogeneous w.r.t. decision labels. This allows for deeper insights on the data and on the decision-making underlying the final user classification. Our framework is able to identify both positive and negative bias resulting in discrimination. We also highlight patterns of discrimination revealed by the homogeneous cluster centroids, which otherwise could not be captured.

Original languageEnglish
Title of host publicationProceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019
Place of PublicationPiscataway
PublisherInstitute of Electrical and Electronics Engineers
Pages187-194
Number of pages8
ISBN (Electronic)978-1-7281-1488-0
DOIs
Publication statusPublished - 1 Jun 2019
Event2nd IEEE International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019 - Cagliari, Sardinia, Italy
Duration: 3 Jun 20195 Jun 2019

Conference

Conference2nd IEEE International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019
CountryItaly
CityCagliari, Sardinia
Period3/06/195/06/19

Fingerprint

Decision making
Learning systems
Labels
Discrimination
Hierarchical clustering
Top-down approach
Industry
Decision-making process

Keywords

  • Bias
  • Clustering
  • Discrimination Discovery
  • KNN
  • Measurements

Cite this

Nasiriani, N., Squicciarini, A., Saldanha, Z., Goel, S., & Zannone, N. (2019). Hierarchical clustering for discrimination discovery: a top-down approach. In Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019 (pp. 187-194). [8791737] Piscataway: Institute of Electrical and Electronics Engineers. https://doi.org/10.1109/AIKE.2019.00041
Nasiriani, Neda ; Squicciarini, Anna ; Saldanha, Zara ; Goel, Sanchit ; Zannone, Nicola. / Hierarchical clustering for discrimination discovery : a top-down approach. Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019. Piscataway : Institute of Electrical and Electronics Engineers, 2019. pp. 187-194
@inproceedings{c4fa29ae40a945b0af82cadcf4d96bb3,
title = "Hierarchical clustering for discrimination discovery: a top-down approach",
abstract = "Today, data is an essential part of many decision-making processes in businesses and social life through the use of various machine learning techniques. These methods can easily perpetuate human bias in the data and result in discrimination. Despite a growing interest in data discrimination discovery and removal, to date there is a lack of a general and robust framework to distinguish discriminatory decision-making processes from non-discriminatory ones. In this work, we present a generic framework that helps detect possible discrimination by analyzing historical data and associated decisions using a top-down unsupervised approach, which we refer to as hierarchical clustering. Our approach is highly adaptive as it gradually 'learns' users' inherent groups, and clusters their records using cohesiveness and density of points in the dataset. Moreover, we propose a progressive attribute-selection method to choose statistically relevant attributes, thus reducing the effect of noise. Finally, we adopt a recursive notion of cluster profile that is homogeneous w.r.t. decision labels. This allows for deeper insights on the data and on the decision-making underlying the final user classification. Our framework is able to identify both positive and negative bias resulting in discrimination. We also highlight patterns of discrimination revealed by the homogeneous cluster centroids, which otherwise could not be captured.",
keywords = "Bias, Clustering, Discrimination Discovery, KNN, Measurements",
author = "Neda Nasiriani and Anna Squicciarini and Zara Saldanha and Sanchit Goel and Nicola Zannone",
year = "2019",
month = "6",
day = "1",
doi = "10.1109/AIKE.2019.00041",
language = "English",
pages = "187--194",
booktitle = "Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019",
publisher = "Institute of Electrical and Electronics Engineers",
address = "United States",

}

Nasiriani, N, Squicciarini, A, Saldanha, Z, Goel, S & Zannone, N 2019, Hierarchical clustering for discrimination discovery: a top-down approach. in Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019., 8791737, Institute of Electrical and Electronics Engineers, Piscataway, pp. 187-194, 2nd IEEE International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019, Cagliari, Sardinia, Italy, 3/06/19. https://doi.org/10.1109/AIKE.2019.00041

Hierarchical clustering for discrimination discovery : a top-down approach. / Nasiriani, Neda; Squicciarini, Anna; Saldanha, Zara; Goel, Sanchit; Zannone, Nicola.

Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019. Piscataway : Institute of Electrical and Electronics Engineers, 2019. p. 187-194 8791737.

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Hierarchical clustering for discrimination discovery

T2 - a top-down approach

AU - Nasiriani, Neda

AU - Squicciarini, Anna

AU - Saldanha, Zara

AU - Goel, Sanchit

AU - Zannone, Nicola

PY - 2019/6/1

Y1 - 2019/6/1

N2 - Today, data is an essential part of many decision-making processes in businesses and social life through the use of various machine learning techniques. These methods can easily perpetuate human bias in the data and result in discrimination. Despite a growing interest in data discrimination discovery and removal, to date there is a lack of a general and robust framework to distinguish discriminatory decision-making processes from non-discriminatory ones. In this work, we present a generic framework that helps detect possible discrimination by analyzing historical data and associated decisions using a top-down unsupervised approach, which we refer to as hierarchical clustering. Our approach is highly adaptive as it gradually 'learns' users' inherent groups, and clusters their records using cohesiveness and density of points in the dataset. Moreover, we propose a progressive attribute-selection method to choose statistically relevant attributes, thus reducing the effect of noise. Finally, we adopt a recursive notion of cluster profile that is homogeneous w.r.t. decision labels. This allows for deeper insights on the data and on the decision-making underlying the final user classification. Our framework is able to identify both positive and negative bias resulting in discrimination. We also highlight patterns of discrimination revealed by the homogeneous cluster centroids, which otherwise could not be captured.

AB - Today, data is an essential part of many decision-making processes in businesses and social life through the use of various machine learning techniques. These methods can easily perpetuate human bias in the data and result in discrimination. Despite a growing interest in data discrimination discovery and removal, to date there is a lack of a general and robust framework to distinguish discriminatory decision-making processes from non-discriminatory ones. In this work, we present a generic framework that helps detect possible discrimination by analyzing historical data and associated decisions using a top-down unsupervised approach, which we refer to as hierarchical clustering. Our approach is highly adaptive as it gradually 'learns' users' inherent groups, and clusters their records using cohesiveness and density of points in the dataset. Moreover, we propose a progressive attribute-selection method to choose statistically relevant attributes, thus reducing the effect of noise. Finally, we adopt a recursive notion of cluster profile that is homogeneous w.r.t. decision labels. This allows for deeper insights on the data and on the decision-making underlying the final user classification. Our framework is able to identify both positive and negative bias resulting in discrimination. We also highlight patterns of discrimination revealed by the homogeneous cluster centroids, which otherwise could not be captured.

KW - Bias

KW - Clustering

KW - Discrimination Discovery

KW - KNN

KW - Measurements

UR - http://www.scopus.com/inward/record.url?scp=85071510303&partnerID=8YFLogxK

U2 - 10.1109/AIKE.2019.00041

DO - 10.1109/AIKE.2019.00041

M3 - Conference contribution

AN - SCOPUS:85071510303

SP - 187

EP - 194

BT - Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019

PB - Institute of Electrical and Electronics Engineers

CY - Piscataway

ER -

Nasiriani N, Squicciarini A, Saldanha Z, Goel S, Zannone N. Hierarchical clustering for discrimination discovery: a top-down approach. In Proceedings - IEEE 2nd International Conference on Artificial Intelligence and Knowledge Engineering, AIKE 2019. Piscataway: Institute of Electrical and Electronics Engineers. 2019. p. 187-194. 8791737 https://doi.org/10.1109/AIKE.2019.00041