Learning Fair Representations for Kernel Models

Fair representations are a powerful tool for satisfying fairness goals such as statistical parity and equality of opportunity in learned models. Existing techniques for learning these representations are typically model-agnostic, as they pre-process the original data such that the output satisfies some fairness criterion, and can be used with arbitrary learning methods. In contrast, we demonstrate the promise of learning a model-aware fair representation, focusing on kernel-based models. We leverage the classical sufficient dimension reduction (SDR) framework to construct representations as subspaces of the reproducing kernel Hilbert space (RKHS), whose member functions are guaranteed to satisfy a given fairness criterion. Our method supports several fairness criteria, continuous and discrete data, and multiple protected attributes. We also characterize the fairness-accuracy trade-off with a parameter that relates to the principal angles between subspaces of the RKHS. Finally, we apply our approach to obtain the first fair Gaussian process (FGP) prior for fair Bayesian learning, and show that it is competitive with, and in some cases outperforms, state-of-the-art methods on real data.