Fairbeat: Assessing and Mitigating Bias with the Composite Balance Score
Abstract
Fairbeat , a novel Fairness Assessment tool for Resampling-based Bias Elimination and Algorithm Training, addresses the critical challenge of fairness in machine learning. Machine learning models often exhibit biases stemming from imbalances in training data concerning protected attributes, leading to discriminatory outcomes. Fairbeat leverages the Composite Balance Score (CBS), a comprehensive metric that evaluates the balance of the dataset by integrating the imbalance of attributes, the imbalance of labels and the association of attributes and labels into a single normalized score. This tool facilitates proactive bias assessment prior to model training, supports multi-class attributes, and provides a user-friendly environment for exploring and visualizing the impact of various bias mitigation techniques, including resampling methods, thereby promoting the development of more equitable and ethically sound AI systems. The demonstration video can be found at https://youtu.be/9aHKfZgtXKg .
Cite
Text
Lequeu et al. "Fairbeat: Assessing and Mitigating Bias with the Composite Balance Score." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2025. doi:10.1007/978-3-032-06129-4_34Markdown
[Lequeu et al. "Fairbeat: Assessing and Mitigating Bias with the Composite Balance Score." European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases, 2025.](https://mlanthology.org/ecmlpkdd/2025/lequeu2025ecmlpkdd-fairbeat/) doi:10.1007/978-3-032-06129-4_34BibTeX
@inproceedings{lequeu2025ecmlpkdd-fairbeat,
title = {{Fairbeat: Assessing and Mitigating Bias with the Composite Balance Score}},
author = {Lequeu, Pierre-Antoine and Lagraa, Sofiane and Robin, Geoffroy and Ouedraogo, Moussa},
booktitle = {European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases},
year = {2025},
pages = {475-480},
doi = {10.1007/978-3-032-06129-4_34},
url = {https://mlanthology.org/ecmlpkdd/2025/lequeu2025ecmlpkdd-fairbeat/}
}