Recovering Structured Data
Recovering data from noise is a fundamental problem that arises in many applications such as compressed sensing, channel communication, pattern matching, and internet advertising. The problem is known to be extremely difficult to solve with no additional information about the datasets. However, efficient recovery may be possible if the datasets have some specific pattern or structure associated with them. In this talk, I will present my work on the algorithmic aspects of recovering certain structured datasets from noise. In particular, I will focus on the data recovery problem in sparse, algebraic, and geometric data sets. Such structured datasets often occur naturally in many applications (for instance, image and speech signals are sparse) or are embedded into data via preprocessing (mapping data points into feature vector space gives it a geometric structure).
Bio: Venkata Gandikota is a postdoctoral researcher at the TRIPODS - Institute for Theoretical Foundations of Data Science at the University of Massachusetts Amherst. Before this, he was a postdoctoral fellow at Johns Hopkins University. He received his Ph.D. from Purdue University in 2017 under the guidance of Prof. Elena Grigorescu. His research interests include coding theory, information theory, and their application to machine learning algorithms.