A systematic review of fairness in machine learning

Fairness in Machine Learning (ML) has emerged as a crucial concern as these models increasingly influence critical decisions in various domains, including healthcare, finance, and criminal justice. The presence of bias in ML systems can lead to unfair and discriminatory outcomes, undermining the reliability and ethical standards of these technologies. As the deployment of ML expands, ensuring that these systems are fair and unbiased is not only a technical challenge but also a moral imperative. Here, a systematic literature review was conducted to explore fairness in machine learning, utilizing the ACM, IEEE, and Springer databases. From an initial retrieval of 975 papers, 30 were included in the review. The results highlight the identification of sensitive attributes, the metrics used to assess bias, and the various databases tested. Additionally, the review categorizes the in-processing and post-processing approaches employed to mitigate bias and examines how studies are managing the trade-off between fairness and accuracy. This comprehensive analysis provides a detailed understanding of the current state of fairness in machine learning and offers insights into effective strategies for bias mitigation.