Supervisors

Suitable for

Abstract

The goal of fair machine learning is to ensure that decisions taken by machine learning systems don't discriminate based on sensitive attributes like race and gender. A limitation/weakness of current approaches to fair machine learning is that they do not account for the behaviour of the humans who use these ML systems as decision-support tools. This may result in fairness being violated in the final decisions. In this project, the student will explore this aspect more formally. In particular, we would be interested in understanding under what conditions, fairness outcomes can improve if the human has more control over the output and internal design of the ML system. The student will be encouraged to use simulations and/or analytical methods to derive the conditions. Depending on time and interest, there will also be an opportunity to conduct human subject experiments.

Prerequisites: Familiarity with machine learning, proficiency in Python, interest in ml fairness and behavioural studies. Students are encouraged to reach out to Naman Goel (naman.goel@cs.ox.ac.uk) to discuss more project ideas related to above topics.