CISE Seminar: Shuchin Aeron, Tufts University

Talk Title: Optimal Transport (OT) and Some of its Applications

Abstract: In this talk we will begin by introducing the well-known problem of Optimal Transport (OT) as a means to compare two distributions and first consider the case of univariate distributions. We apply the resulting theory towards developing a non-parameteric and unsupervised detection of change points in time series data. In this context we exploit recent results that utilize the OT framework for non-parametric two sample testing and propose a matched filtering approach to smooth the noisy test statistics. Results on real data show the efficiency of the proposed approach in segmenting time series data. If time permits, we will introduce the theory of OT for multivariate distributions and discuss our current research efforts that exploit it towards addressing statistical and machine learning problems.

Shuchin Aeron is an Associate Professor in the Dept. of ECE at Tufts University. He received his B. Tech from IIT Kanpur in 2002, MS from Boston University in 2004, and PhD from Boston University in 2009, all in Electrical Engineering. Prior to joining Tufts University in 2011, he was a post-doctoral fellow at Schlumberger Doll Research from 2009-2011. He received the NSF CAREER award in 2016. He is currently a senior member of the IEEE, an associate editor for IEEE Transactions on Geosciences and Remote Sensing, and is on the technical committee of Machine Learning for Signal Processing, IEEE SPS. He is also the co-director of the Data Science programs (MS and BS) in the School of Engineering at Tufts University. Prof. Aeron’s research interests lie in Information Theory, Tensor Data Analytics, and more recently in Mathematical Statistics and Optimal Transport.

Faculty Host: Prakash Ishwar<br>

Student Host: Mahroo Bahreinian

When 3:00 pm to 4:30 pm on Friday, April 16, 2021
Location Virtual Event (Register for login information.)