Data vs. Model Machine Learning Fairness Testing: An Empirical Study

Shome, A.; Cruz, Luis; van Deursen, A.

doi:10.1145/3639478.3643121

Data vs. Model Machine Learning Fairness Testing

Title

Data vs. Model Machine Learning Fairness Testing: An Empirical Study

Author

Shome, A. (TU Delft Software Engineering)
Cruz, Luis (TU Delft Software Engineering)
van Deursen, A. (TU Delft Software Engineering)

Date

2024

Abstract

Although several fairness definitions and bias mitigation techniques exist in the literature, all existing solutions evaluate fairness of Machine Learning (ML) systems after the training stage. In this paper, we take the first steps towards evaluating a more holistic approach by testing for fairness both before and after model training. We evaluate the effectiveness of the proposed approach and position it within the ML development lifecycle, using an empirical analysis of the relationship between model dependent and independent fairness metrics. The study uses 2 fairness metrics, 4 ML algorithms, 5 real-world datasets and 1600 fairness evaluation cycles. We find a linear relationship between data and model fairness metrics when the distribution and the size of the training data changes. Our results indicate that testing for fairness prior to training can be a "cheap" and effective means of catching a biased data collection process early; detecting data drifts in production systems and minimising execution of full training cycles thus reducing development time and costs.

Subject

Datacentric AI
Empirical Software Engineering
ML Fairness Testing
SE4ML

To reference this document use:

http://resolver.tudelft.nl/uuid:b6618a7d-78b0-44b7-9dc1-b0a2455595cf

DOI

https://doi.org/10.1145/3639478.3643121

Publisher

IEEE

ISBN

9798400705021

Source

Proceedings - 2024 ACM/IEEE 46th International Conference on Software Engineering: Companion, ICSE-Companion 2024