Keynote Speaker

Rich Caruana, senior principal researcher at Microsoft Research

Title: Friends Don’t Let Friends Deploy Black-Box Models: The Importance of Intelligbility and Explanation for Machine Learning in Healthcare

Abstract: In machine learning sometimes a tradeoff must be made between accuracy and intelligibility: the most accurate models usually are often not very intelligible, and the most intelligible models usually are less accurate.  This can limit the accuracy of models that can safely be deployed in mission-critical applications such as healthcare where being able to understand, validate, edit, and ultimately trust a model is important.  We have developed a learning method based on generalized additive models (GAMs) that is as accurate as full complexity models such as boosted trees and random forests, but even more intelligible than linear models.  This makes it easy to understand what models have learned and to edit models when they learn inappropriate things.  Making it possible for medical experts to understand and repair a model is critical because most clinical data is complex and has unanticipated problems.  I'll present a number of healthcare case studies where these high-accuracy GAMs discover surprising patterns in the data that would have made deploying black-box models risky. The case studies include surprising findings in pregnancy, pneumonia, ICU and COVID-19 risk prediction.

Bio: Rich Caruana is a Senior Principal Researcher at Microsoft. His research focus is on intelligible/transparent modeling and machine learning for medical decision making. Before joining Microsoft, Rich was on the faculty at Cornell, at UCLA's Medical School, and at CMU's Center for Learning and Discovery. Rich's Ph.D. is from CMU, and his thesis on Multitask Learning helped create interest in the new subfield of Transfer Learning. Rich has received a number of best paper awards, an NSF CAREER Award in 2004 for Meta Clustering, and chaired KDD in 2007.