Avoiding Discrimination in Automated Decision Making and Machine Learning
As automatic systems increasingly make decisions affecting humans, we must understand how our notions of fairness can be incorporated in a principled way. We aim to implement various recent methods for discrimination discovery and removal in data as well as inherently non-discriminatory algorithms. An easy to use library will be made openly available to audit data sets or algorithms and scrutinize their fairness. A detailed documentation and guidelines on how to interpret the results should render it a useful tool for a wide audience.
We are hiring!
We are currently looking for a Master's student or student research assistant to help us with this project.
- Starting Date: anytime after March 20, 2018
- Duration: negotiable between 6 and 12 months
- Location: Max Planck Institute for Intelligent Systems, Empirical Inference, Max Planck Ring 4, 72076 Tübingen
- You will get a desk at the office.
- Working remotely is possible.
- Payment: Depending on working hours up to 640€ / month
- Supervision: Niki Kilbertus (MPI & University of Cambridge), Bernhard Schölkopf (MPI & Amazon) and Adrian Weller (University of Cambridge)
- Application: Send a short description of your background and motivation to email@example.com
During the project you will familiarize yourself with the relevant literature before we jointly evaluate previously proposed methods in terms of applicability. This includes reading, understanding and replicating application oriented papers in the field of fair learning. In addition to this research component, the major focus lies on the design and implementation of an easy to use open source library in Python with extensive documentation and usage guidelines. The project is funded by a Digital Impact Grant from the Digital Civil Society Lab at Stanford University and was initiated by Bernhard Schölkopf, Adrian Weller and myself.
This project is a unique opportunity to interact and work with many bright and experienced researchers at an internationally renowned machine learning lab. Previous Master students and research assistants often went on to do machine learning PhDs at top tier schools. It will provide you with a working knowledge of the current state of the art in algorithmic fairness, a field of growing importance and interest.
For an introduction to fairness in machine learning see the slides of a tutorial at NIPS 2017 by Solon Barocas and Moritz Hardt. A first literature overview can be found on the course homepage of Moritz Hardt's course on Fairness in Machine Learning.
Who are we looking for
You are a keen student with strong programming skills (preferably in Python) and a solid math background (computer science, mathematics, physics or similar). You have a deep interest in machine learning research and are keen to learn about algorithmic fairness by taking apart current research papers. You are attentive to details and write well documented code.
Please send a short description of your background and motivation to firstname.lastname@example.org so we can discuss the details.
Image Credit: Moritz Hardt (https://fairmlclass.github.io/1.html#/4)