Subgroup Harm Assessor: Identifying Potential Fairness-Related Harms and Predictive Bias

Adam Dubowski; Hilde Weerts; Anouk Wolters; Mykola Pechenizkiy

doi:10.1007/978-3-031-70371-3_31

Subgroup Harm Assessor: Identifying Potential Fairness-Related Harms and Predictive Bias

Adam Dubowski, Hilde Weerts, Anouk Wolters, Mykola Pechenizkiy

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

Abstract

With the integration of artificial intelligence into real-world decision-support systems, there is increasing interest in tools that facilitate the identification of potential biases and fairness-related harms of machine learning models. While existing toolkits provide approaches to evaluate harms associated with discrete predicted outcomes, the assessment of disparities in epistemic value provided by continuous risk scores is relatively underexplored. Additionally, relatively few works focus on identifying the biases at the root of the harm. In this work, we present a visual analytics “Subgroup Harm Assessor” tool that allows users to: (1) identify disparities in the epistemic value of risk-scoring models via subgroup discovery of disparities in model log loss, (2) evaluate the extent to which the disparity might be caused by disparities in the informativeness of features via SHapley Additive exPlanations (SHAP) of model loss.

Original language	English
Title of host publication	Machine Learning and Knowledge Discovery in Databases. Research Track and Demo Track
Subtitle of host publication	European Conference, ECML PKDD 2024, Vilnius, Lithuania, September 9–13, 2024, Proceedings, Part VIII
Editors	Albert Bifet, Povilas Daniušis, Jesse Davis, Tomas Krilavičius, Meelis Kull, Eirini Ntoutsi, Kai Puolamäki, Indrė Žliobaitė
Place of Publication	Cham
Publisher	Springer
Pages	413-417
Number of pages	5
ISBN (Electronic)	978-3-031-70371-3
ISBN (Print)	978-3-031-70370-6
DOIs	https://doi.org/10.1007/978-3-031-70371-3_31
Publication status	Published - 22 Aug 2024
Event	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2024 - Vilnius, Lithuania Duration: 9 Sept 2024 → 13 Sept 2024

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	14948 LNAI
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2024
Country/Territory	Lithuania
City	Vilnius
Period	9/09/24 → 13/09/24

Keywords

AI
Explainability
Fairness
risk
XAI

Access to Document

10.1007/978-3-031-70371-3_31

Cite this

Dubowski, A., Weerts, H., Wolters, A., & Pechenizkiy, M. (2024). Subgroup Harm Assessor: Identifying Potential Fairness-Related Harms and Predictive Bias. In A. Bifet, P. Daniušis, J. Davis, T. Krilavičius, M. Kull, E. Ntoutsi, K. Puolamäki, & I. Žliobaitė (Eds.), Machine Learning and Knowledge Discovery in Databases. Research Track and Demo Track: European Conference, ECML PKDD 2024, Vilnius, Lithuania, September 9–13, 2024, Proceedings, Part VIII (pp. 413-417). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14948 LNAI). Springer. https://doi.org/10.1007/978-3-031-70371-3_31

Dubowski, Adam ; Weerts, Hilde ; Wolters, Anouk et al. / Subgroup Harm Assessor : Identifying Potential Fairness-Related Harms and Predictive Bias. Machine Learning and Knowledge Discovery in Databases. Research Track and Demo Track: European Conference, ECML PKDD 2024, Vilnius, Lithuania, September 9–13, 2024, Proceedings, Part VIII. editor / Albert Bifet ; Povilas Daniušis ; Jesse Davis ; Tomas Krilavičius ; Meelis Kull ; Eirini Ntoutsi ; Kai Puolamäki ; Indrė Žliobaitė. Cham : Springer, 2024. pp. 413-417 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{4999b438b4a0476497f8945e026d0646,

title = "Subgroup Harm Assessor: Identifying Potential Fairness-Related Harms and Predictive Bias",

abstract = "With the integration of artificial intelligence into real-world decision-support systems, there is increasing interest in tools that facilitate the identification of potential biases and fairness-related harms of machine learning models. While existing toolkits provide approaches to evaluate harms associated with discrete predicted outcomes, the assessment of disparities in epistemic value provided by continuous risk scores is relatively underexplored. Additionally, relatively few works focus on identifying the biases at the root of the harm. In this work, we present a visual analytics “Subgroup Harm Assessor” tool that allows users to: (1) identify disparities in the epistemic value of risk-scoring models via subgroup discovery of disparities in model log loss, (2) evaluate the extent to which the disparity might be caused by disparities in the informativeness of features via SHapley Additive exPlanations (SHAP) of model loss.",

keywords = "AI, Explainability, Fairness, risk, XAI",

author = "Adam Dubowski and Hilde Weerts and Anouk Wolters and Mykola Pechenizkiy",

year = "2024",

month = aug,

day = "22",

doi = "10.1007/978-3-031-70371-3_31",

language = "English",

isbn = "978-3-031-70370-6",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "413--417",

editor = "Albert Bifet and Povilas Daniu{\v s}is and Jesse Davis and Tomas Krilavi{\v c}ius and Meelis Kull and Eirini Ntoutsi and Kai Puolam{\"a}ki and Indrė {\v Z}liobaitė",

booktitle = "Machine Learning and Knowledge Discovery in Databases. Research Track and Demo Track",

address = "Germany",

note = "European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2024 ; Conference date: 09-09-2024 Through 13-09-2024",

}

Dubowski, A, Weerts, H, Wolters, A & Pechenizkiy, M 2024, Subgroup Harm Assessor: Identifying Potential Fairness-Related Harms and Predictive Bias. in A Bifet, P Daniušis, J Davis, T Krilavičius, M Kull, E Ntoutsi, K Puolamäki & I Žliobaitė (eds), Machine Learning and Knowledge Discovery in Databases. Research Track and Demo Track: European Conference, ECML PKDD 2024, Vilnius, Lithuania, September 9–13, 2024, Proceedings, Part VIII. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14948 LNAI, Springer, Cham, pp. 413-417, European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2024, Vilnius, Lithuania, 9/09/24. https://doi.org/10.1007/978-3-031-70371-3_31

Subgroup Harm Assessor: Identifying Potential Fairness-Related Harms and Predictive Bias. / Dubowski, Adam; Weerts, Hilde; Wolters, Anouk et al.
Machine Learning and Knowledge Discovery in Databases. Research Track and Demo Track: European Conference, ECML PKDD 2024, Vilnius, Lithuania, September 9–13, 2024, Proceedings, Part VIII. ed. / Albert Bifet; Povilas Daniušis; Jesse Davis; Tomas Krilavičius; Meelis Kull; Eirini Ntoutsi; Kai Puolamäki; Indrė Žliobaitė. Cham: Springer, 2024. p. 413-417 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 14948 LNAI).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Subgroup Harm Assessor

T2 - European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, ECML PKDD 2024

AU - Dubowski, Adam

AU - Weerts, Hilde

AU - Wolters, Anouk

AU - Pechenizkiy, Mykola

PY - 2024/8/22

Y1 - 2024/8/22

N2 - With the integration of artificial intelligence into real-world decision-support systems, there is increasing interest in tools that facilitate the identification of potential biases and fairness-related harms of machine learning models. While existing toolkits provide approaches to evaluate harms associated with discrete predicted outcomes, the assessment of disparities in epistemic value provided by continuous risk scores is relatively underexplored. Additionally, relatively few works focus on identifying the biases at the root of the harm. In this work, we present a visual analytics “Subgroup Harm Assessor” tool that allows users to: (1) identify disparities in the epistemic value of risk-scoring models via subgroup discovery of disparities in model log loss, (2) evaluate the extent to which the disparity might be caused by disparities in the informativeness of features via SHapley Additive exPlanations (SHAP) of model loss.

AB - With the integration of artificial intelligence into real-world decision-support systems, there is increasing interest in tools that facilitate the identification of potential biases and fairness-related harms of machine learning models. While existing toolkits provide approaches to evaluate harms associated with discrete predicted outcomes, the assessment of disparities in epistemic value provided by continuous risk scores is relatively underexplored. Additionally, relatively few works focus on identifying the biases at the root of the harm. In this work, we present a visual analytics “Subgroup Harm Assessor” tool that allows users to: (1) identify disparities in the epistemic value of risk-scoring models via subgroup discovery of disparities in model log loss, (2) evaluate the extent to which the disparity might be caused by disparities in the informativeness of features via SHapley Additive exPlanations (SHAP) of model loss.

KW - AI

KW - Explainability

KW - Fairness

KW - risk

KW - XAI

UR - http://www.scopus.com/inward/record.url?scp=85203876265&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-70371-3_31

DO - 10.1007/978-3-031-70371-3_31

M3 - Conference contribution

AN - SCOPUS:85203876265

SN - 978-3-031-70370-6

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 413

EP - 417

BT - Machine Learning and Knowledge Discovery in Databases. Research Track and Demo Track

A2 - Bifet, Albert

A2 - Daniušis, Povilas

A2 - Davis, Jesse

A2 - Krilavičius, Tomas

A2 - Kull, Meelis

A2 - Ntoutsi, Eirini

A2 - Puolamäki, Kai

A2 - Žliobaitė, Indrė

PB - Springer

CY - Cham

Y2 - 9 September 2024 through 13 September 2024

ER -

Dubowski A, Weerts H, Wolters A, Pechenizkiy M. Subgroup Harm Assessor: Identifying Potential Fairness-Related Harms and Predictive Bias. In Bifet A, Daniušis P, Davis J, Krilavičius T, Kull M, Ntoutsi E, Puolamäki K, Žliobaitė I, editors, Machine Learning and Knowledge Discovery in Databases. Research Track and Demo Track: European Conference, ECML PKDD 2024, Vilnius, Lithuania, September 9–13, 2024, Proceedings, Part VIII. Cham: Springer. 2024. p. 413-417. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-70371-3_31