Presenter: Dr. Praneeth Narayanamurthy
Title: Provable and Efficient Algorithms for Robust Subspace Learning and Tracking
Abstract: In today's big-data age, an enormous amount of high-dimensional data is being generated, transmitted, processed, and stored. Although very high-dimensional, the data often possesses an underlying low-dimensional representation, and for time-series data, these representations typically change with time. In this talk, we will explore the problem of learning and tracking a slowly changing low-dimensional linear subspace from corrupted data. In particular we explore the cases where pats of the data is missing, and also where the data is corrupted by gross (sparse) outliers. We design and analyze provable, fast, and memory efficient algorithms to recover the underlying linear subspace(s). We also provide extensive empirical evaluation of these algorithms on the video layering experiment.
Bio: Praneeth Narayanamurthy is a Ph.D. student in the Department of Electrical and Computer Engineering at Iowa State University. He previously obtained his B.Tech degree in Electrical and Electronics Engineering from National Institute of Technology Karnataka in 2014. His research interests include the algorithmic and theoretical aspects of High-Dimensional Statistical Signal Processing, and Machine Learning.
After the presentation, there will be a short time for discussion and questions afterwards.