IDEaS Short Talks and Networking Social
“Machine Learning’s Unequal Representation” by Jamie Morgenstern
Monday, October 22nd
Part of Two Short 30-MinuteTalks from 2-3 pm
Networking Social from 3-4 pm
Technology Square Research Building (TSRB) Auditorium
(You are welcome to attend any part of these events as your schedule permits.)
IDEaS is running a series of short talks to learn about research across the Georgia Tech campus. The presentations are from broadly different topics and accessible to those in other research areas.
Assistant Professor in the School of Computer Science
Georgia Institute of Technology
2:00 - 2:30 pm
“Machine Learning’s Unequal Representation”
Recent news coverage of machine learning systems has included numerous examples of ML systems’ differing behavior towards different demographic groups: image searches for CEO return male images at higher rates than men inhabit those roles, minority profiles are more likely to be shown DUI help advertisements, and facial recognition systems often have higher error rates on dark-skinned women. From a technical perspective, there may be many sources of this behavior. I will outline a few of these possible sources of unfairness, and describe some recent technical progress on addressing them.
Jamie Morgenstern is an assistant professor in the School of Computer Science Georgia Tech. Prior to this appointment, she was hosted by Michael Kearns, Aaron Roth, and Rakesh Vohra as a Warren Center fellow at the University of Pennsylvania. She completed her PhD at Carnegie Mellon University in 2015 under the advisement of Avrim Blum. She studies the social impact of machine learning and the impact of social behavior on ML's guarantees. How should machine learning be made robust to behavior of the people generating training or test data for it? How should ensure that the models we design do not exacerbate inequalities already present in society?