Matthew Joseph

ma + jos + at + cis.upenn.edu

3401 Walnut St., 416-B

I'm a fourth-year PhD student in the Theory Group at UPenn, advised by Aaron Roth. Here's my CV.

Research

I'm interested in theoretical tools that help us reason about the social dimensions of machine learning in a rigorous way. What does it mean for machine learning to be privacy-preserving or fair, and how can we build algorithms that meet those goals?

Papers

All papers are alphabetical author order.

  1. Local Differential Privacy for Evolving Data (NIPS 2018, Spotlight)
    With Aaron Roth, Jonathan Ullman, and Bo Waggoner.

  2. Meritocratic Fairness for Infinite and Contextual Bandits (AIES 2018, Poster)
    With Michael Kearns, Jamie Morgenstern, Seth Neel, and Aaron Roth.

  3. A Convex Framework for Fair Regression (FATML 2017, Poster)
    With Richard Berk, Hoda Heidari, Shahin Jabbari, Michael Kearns, Jamie Morgenstern, Seth Neel, and Aaron Roth.

  4. Fairness in Reinforcement Learning (ICML 2017, Talk)
    With Shahin Jabbari, Michael Kearns, Jamie Morgenstern, and Aaron Roth.

  5. Fairness in Learning: Classic and Contextual Bandits (NIPS 2016, Poster)
    With Michael Kearns, Jamie Morgenstern, and Aaron Roth.

Other

I also enjoy cooking and falling off things.