Abstract

With the wide application of machine learning driven automated decisions (i.e., education, loan approval and hiring) in daily life, it is critical to address the problem of discriminatory behaviour toward certain individuals or groups. Early studies focus on defining the correlation/association-based notions, such as statistical parity, equalized odds and so on. But recent researches reflect that it is necessary to use causality to address the problem of fairness. This review provide an exhaustive overview of notions and methods for detecting and eliminating algorithmic discrimination from a causality perspective. The review begin by introducing the common causality-based definitions and measures for fairness. It then review causality-based fairness-enhancing methods from the perspective of pre-processing, in-processing and post-processing mechanisms, and conduct a comprehensive analysis of the advantages, disadvantages and applicability of these mechanisms. In addition to that, this review also examine other domains, where researchers have observed unfair outcomes and ways they have tried to address them. There are still many challenges that hinder the practical application of causality-based fairness notions, specifically the difficulty of acquiring causal graphs and identifiability of causal effects. One of the main purposes of this review is to spark more researchers to tackle these challenges in the near future.