Achieving non-discrimination in prediction

TL;DR

This paper provides a theoretical framework to ensure non-discrimination in predictions, revealing limitations of pre-process methods and proposing a new two-phase approach with guarantees.

Contribution

It introduces a formal causal model for discrimination, proves that removing training data discrimination doesn't guarantee fair predictions, and develops a framework with theoretical guarantees.

Findings

Discrimination in prediction can persist even if training data is non-discriminatory.

Not all pre-process methods guarantee non-discrimination in predictions.

The proposed two-phase framework effectively ensures non-discrimination with theoretical backing.

Abstract

Discrimination-aware classification is receiving an increasing attention in data science fields. The pre-process methods for constructing a discrimination-free classifier first remove discrimination from the training data, and then learn the classifier from the cleaned data. However, they lack a theoretical guarantee for the potential discrimination when the classifier is deployed for prediction. In this paper, we fill this gap by mathematically bounding the probability of the discrimination in prediction being within a given interval in terms of the training data and classifier. We adopt the causal model for modeling the data generation mechanism, and formally defining discrimination in population, in a dataset, and in prediction. We obtain two important theoretical results: (1) the discrimination in prediction can still exist even if the discrimination in the training data is completely removed; and (2) not all pre-process methods can ensure non-discrimination in prediction even though they can achieve non-discrimination in the modified training data. Based on the results, we develop a two-phase framework for constructing a discrimination-free classifier with a theoretical guarantee. The experiments demonstrate the theoretical results and show the effectiveness of our two-phase framework.