Optimized Pre-Processing for Discrimination Prevention

TLDR

Non-discrimination is a recognized objective in algorithmic decision making. The paper introduces a novel probabilistic formulation of data pre-processing to reduce discrimination. The authors formulate a convex optimization that learns a data transformation balancing discrimination control, minimal distortion of individual samples, and utility preservation, analyze sample‑size effects, and apply two instances to synthetic and real criminal recidivism data. Results show discrimination can be greatly reduced with only a small loss in classification accuracy.

Abstract

Non-discrimination is a recognized objective in algorithmic decision making. In this paper, we introduce a novel probabilistic formulation of data pre-processing for reducing discrimination. We propose a convex optimization for learning a data transformation with three goals: controlling discrimination, limiting distortion in individual data samples, and preserving utility. We characterize the impact of limited sample size in accomplishing this objective. Two instances of the proposed optimization are applied to datasets, including one on real-world criminal recidivism. Results show that discrimination can be greatly reduced at a small cost in classification accuracy.