Law & Economics Workshop
presents
Cynthia Rudin
Duke University Computer Science
Transparency, causality, and bias:
Can we really use machine learning for high stakes decisions?
Tuesday, September 26, 2017
4:10 – 5:45
Jackman Law Building, Rm. J230
78 Queen’s Park
Standard machine learning techniques are problematic when used for high-stakes decisions, such as those arising in criminal justice. One serious problem is lack of transparency in the models, which causes arguments over whether the models are racially biased. "Black box" models also cause problems when data are entered incorrectly, causing misleading predictions that are difficult to detect. The U.S. Justice System currently suffers from these issues, since proprietary risk prediction tools are used for decisions about parole, sentencing, and other applications. There have been cases in the U.S. where prisoners have been denied parole due to incorrect data entered about them. Besides lack of transparency, a separate problem with standard machine learning methods is that they are not usually designed for causal inference. People often tend to misinterpret models as causal when in fact they cannot be interpreted that way. I will argue that these are not inherently problems with machine learning as a field, but instead, the problems stem from using the wrong machine learning techniques. In the first half of the talk, I will present machine learning techniques that create interpretable predictive models, and present applications to recidivism prediction in the U.S. In the second half of the talk I will discuss whether machine learning methods can be created to be appropriate for some specific causal inference settings.
Cynthia Rudin is an associate professor of computer science, electrical and computer engineering, and statistics at Duke University, and directs the Prediction Analysis Lab. Her interests are in machine learning, data mining, applied statistics, and knowledge discovery (Big Data). Her application areas are in energy grid reliability, healthcare, and computational criminology. Previously, Prof. Rudin held positions at MIT, Columbia, and NYU. She holds an undergraduate degree from the University at Buffalo, and a PhD in applied and computational mathematics from Princeton University. She is the recipient of the 2013 and 2016 INFORMS Innovative Applications in Analytics Awards, an NSF CAREER award, was named as one of the "Top 40 Under 40" by Poets and Quants in 2015, and was named by Business insider.com as one of the 12 most impressive professors at MIT in 2015. Work from her lab has won 10 best paper awards in the last 5 years. She is past chair of the INFORMS Data Mining Section, and is currently chair of the Statistical Learning and Data Science section of the American Statistical Association. She also serves on (or has served on) committees for DARPA, the National Institute of Justice, and the National Academy of Sciences (for both statistics and criminology/law).
For more workshop information, please contact Nadia Gulezko at n.gulezko@utoronto.ca.
