Primary Submission Category: Machine Learning and Causal Inference
Statistical Learning for Constrained Functional Parameters in Infinite-Dimensional Models with Applications in Fair Machine Learning
Authors: Razieh Nabi, Nima Hejazi, Mark van der Laan, David Benkeser,
Presenting Author: Razieh Nabi*
Statistical machine learning algorithms, integral to sectors like hiring, finance, and healthcare, risk reinforcing societal biases based on gender, race, religion, among others. To combat this, it’s vital to design models adhering to fairness norms. This involves embedding fairness constraints such as ‘equal opportunity’, ensuring uniform true positive rates across groups, and ‘counterfactual fairness’, assessing outcomes in varied hypothetical scenarios. This study doesn’t favor a specific fairness criterion but proposes a general framework for deriving optimal prediction functions under various constraints. It conceptualizes the learning problem as estimating a constrained functional parameter within a comprehensive statistical model, using a Lagrange-type penalty. This enables representing a fair prediction function in relation to an unfair counterpart, plus other parameters, for integration with standard learning frameworks. Key contributions of our work include a flexible framework for solving constrained optimization problems, closed-form solutions for specific fairness constraints, and an algorithm-neutral approach to fair learning. This framework’s applicability extends beyond algorithmic fairness to other constrained learning contexts like Neyman-Pearson classification, churn reduction, adversarial learning, and reinforcement learning, demonstrating broad utility and impact.