Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps

Isel Grau; Gonzalo Nápoles; Fabian Hoitsma; Lisa Koutsoviti Koumeri; Koen Vanhoof

doi:10.1007/978-3-031-47721-8_50

Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps

Isel Grau, Gonzalo Nápoles, Fabian Hoitsma, Lisa Koutsoviti Koumeri, Koen Vanhoof

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

1 Citation (Scopus)

Abstract

In this paper, we integrate the concepts of feature importance with implicit bias in the context of pattern classification. This is done by means of a three-step methodology that involves (i) building a classifier and tuning its hyperparameters, (ii) building a Fuzzy Cognitive Map model able to quantify implicit bias, and (iii) using the SHAP feature importance to active the neural concepts when performing simulations. The results using a real case study concerning fairness research support our two-fold hypothesis. On the one hand, it is illustrated the risks of using a feature importance method as an absolute tool to measure implicit bias. On the other hand, it is concluded that the amount of bias towards protected features might differ depending on whether the features are numerically or categorically encoded.

Original language	English
Title of host publication	Intelligent Systems and Applications
Subtitle of host publication	Proceedings of the 2023 Intelligent Systems Conference (IntelliSys) Volume 1
Editors	Kohei Arai
Place of Publication	Cham
Publisher	Springer
Pages	745-764
Number of pages	20
ISBN (Electronic)	978-3-031-47721-8
ISBN (Print)	978-3-031-47720-1
DOIs	https://doi.org/10.1007/978-3-031-47721-8_50
Publication status	Published - 10 Jan 2024

Publication series

Name	Lecture Notes in Networks and Systems (LNNS)
Volume	822
ISSN (Print)	2367-3370
ISSN (Electronic)	2367-3389

Keywords

Explainable artificial intelligence
Fairness
Feature importance
Fuzzy cognitive maps
Implicit Bias

Access to Document

10.1007/978-3-031-47721-8_50

Measuring_implicit_bias_using_SHAP_feature_importance_and_Fuzzy_Cognitive_MapsFinal author version , 605 KBLicence: TAVERNE
pdfFinal published version, 862 KBLicence: TAVERNE

Cite this

Grau, I., Nápoles, G., Hoitsma, F., Koumeri, L. K., & Vanhoof, K. (2024). Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps. In K. Arai (Ed.), Intelligent Systems and Applications: Proceedings of the 2023 Intelligent Systems Conference (IntelliSys) Volume 1 (pp. 745-764). (Lecture Notes in Networks and Systems (LNNS); Vol. 822). Springer. https://doi.org/10.1007/978-3-031-47721-8_50

@inproceedings{8b7f7b0d8e5743898047e5be56e3f2cc,

title = "Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps",

abstract = "In this paper, we integrate the concepts of feature importance with implicit bias in the context of pattern classification. This is done by means of a three-step methodology that involves (i) building a classifier and tuning its hyperparameters, (ii) building a Fuzzy Cognitive Map model able to quantify implicit bias, and (iii) using the SHAP feature importance to active the neural concepts when performing simulations. The results using a real case study concerning fairness research support our two-fold hypothesis. On the one hand, it is illustrated the risks of using a feature importance method as an absolute tool to measure implicit bias. On the other hand, it is concluded that the amount of bias towards protected features might differ depending on whether the features are numerically or categorically encoded.",

keywords = "Explainable artificial intelligence, Fairness, Feature importance, Fuzzy cognitive maps, Implicit Bias",

author = "Isel Grau and Gonzalo N{\'a}poles and Fabian Hoitsma and Koumeri, {Lisa Koutsoviti} and Koen Vanhoof",

year = "2024",

month = jan,

day = "10",

doi = "10.1007/978-3-031-47721-8_50",

language = "English",

isbn = "978-3-031-47720-1",

series = "Lecture Notes in Networks and Systems (LNNS)",

publisher = "Springer",

pages = "745--764",

editor = "Kohei Arai",

booktitle = "Intelligent Systems and Applications",

address = "Germany",

}

Grau, I, Nápoles, G, Hoitsma, F, Koumeri, LK & Vanhoof, K 2024, Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps. in K Arai (ed.), Intelligent Systems and Applications: Proceedings of the 2023 Intelligent Systems Conference (IntelliSys) Volume 1. Lecture Notes in Networks and Systems (LNNS), vol. 822, Springer, Cham, pp. 745-764. https://doi.org/10.1007/978-3-031-47721-8_50

Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps. / Grau, Isel; Nápoles, Gonzalo; Hoitsma, Fabian et al.
Intelligent Systems and Applications: Proceedings of the 2023 Intelligent Systems Conference (IntelliSys) Volume 1. ed. / Kohei Arai. Cham: Springer, 2024. p. 745-764 (Lecture Notes in Networks and Systems (LNNS); Vol. 822).

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

TY - GEN

T1 - Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps

AU - Grau, Isel

AU - Nápoles, Gonzalo

AU - Hoitsma, Fabian

AU - Koumeri, Lisa Koutsoviti

AU - Vanhoof, Koen

PY - 2024/1/10

Y1 - 2024/1/10

N2 - In this paper, we integrate the concepts of feature importance with implicit bias in the context of pattern classification. This is done by means of a three-step methodology that involves (i) building a classifier and tuning its hyperparameters, (ii) building a Fuzzy Cognitive Map model able to quantify implicit bias, and (iii) using the SHAP feature importance to active the neural concepts when performing simulations. The results using a real case study concerning fairness research support our two-fold hypothesis. On the one hand, it is illustrated the risks of using a feature importance method as an absolute tool to measure implicit bias. On the other hand, it is concluded that the amount of bias towards protected features might differ depending on whether the features are numerically or categorically encoded.

AB - In this paper, we integrate the concepts of feature importance with implicit bias in the context of pattern classification. This is done by means of a three-step methodology that involves (i) building a classifier and tuning its hyperparameters, (ii) building a Fuzzy Cognitive Map model able to quantify implicit bias, and (iii) using the SHAP feature importance to active the neural concepts when performing simulations. The results using a real case study concerning fairness research support our two-fold hypothesis. On the one hand, it is illustrated the risks of using a feature importance method as an absolute tool to measure implicit bias. On the other hand, it is concluded that the amount of bias towards protected features might differ depending on whether the features are numerically or categorically encoded.

KW - Explainable artificial intelligence

KW - Fairness

KW - Feature importance

KW - Fuzzy cognitive maps

KW - Implicit Bias

UR - http://www.scopus.com/inward/record.url?scp=85182512259&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-47721-8_50

DO - 10.1007/978-3-031-47721-8_50

M3 - Conference contribution

SN - 978-3-031-47720-1

T3 - Lecture Notes in Networks and Systems (LNNS)

SP - 745

EP - 764

BT - Intelligent Systems and Applications

A2 - Arai, Kohei

PB - Springer

CY - Cham

ER -

Grau I, Nápoles G, Hoitsma F, Koumeri LK, Vanhoof K. Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps. In Arai K, editor, Intelligent Systems and Applications: Proceedings of the 2023 Intelligent Systems Conference (IntelliSys) Volume 1. Cham: Springer. 2024. p. 745-764. (Lecture Notes in Networks and Systems (LNNS)). doi: 10.1007/978-3-031-47721-8_50

Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps

Abstract

Publication series

Keywords

Access to Document

Other files and links

Fingerprint

Measuring Implicit Bias Using SHAP Feature Importance and Fuzzy Cognitive Maps

Code: Implicit bias with FCMs

Modeling implicit bias with fuzzy cognitive maps

Cite this