Statistical Fairness

Statistical fairness in machine learning aims to mitigate biases in algorithms that lead to discriminatory outcomes for certain demographic groups. Current research focuses on developing and comparing various fairness metrics, investigating the relationships between different fairness notions (e.g., statistical vs. causal fairness), and designing methods to achieve fairness in diverse model architectures, including diffusion models and federated learning systems, often through techniques like data reweighing or counterfactual data augmentation. This field is crucial for ensuring ethical and equitable use of AI in high-stakes applications, impacting areas such as loan applications, healthcare, and criminal justice.