CISE Seminar: Clayton Scott, Professor, University of Michigan

Friday, March 18, 2022
3:00-4:00pm
Location: Virtual

Clustering from Paired Observations

Clustering is a fundamental yet notoriously difficult machine learning problem. One approach to overcoming the challenges posed by clustering is to assume access to some sort of side information about the clusters. In this talk, I will address the setting where the data to be clustered come in pairs, meaning observations in the same pair belong to the same (unknown) cluster. Adopting a mixture modeling framework, I will argue that the components of a mixture model can be inferred from paired observations under very weak assumptions on the mixture components (clusters). In particular, the clusters may overlap substantially and have nonparametric forms. An application to topic modelling will also be discussed.

Clayton Scott received his PhD in Electrical Engineering from Rice University in 2004, and is currently Professor of Electrical Engineering and Computer Science at the University of Michigan. His research interests include statistical machine learning theory and algorithms, with an emphasis on nonparametric methods for supervised and unsupervised learning. He has also worked on a number of applications including brain imaging, nuclear threat detection, environmental monitoring, and computational biology. In 2010, he received the Career Award from the National Science Foundation.

Faculty Host: Venkatesh Saligrama
Student Host: Andres Chavez Armijos