On Representing Useful Information: Embeddings, Quantization, and Distributed Coding
Low-dimensional embeddings have recently emerged as a key component in modern signal processing theory and practice. This talk will explore the information-preserving properties of such embeddings, which make them suitable for a number of signal representation applications. In particular, we will demonstrate that embeddings can be designed to preserve different ranges of distances with different accuracy. Such designs make embeddings a very convenient coding mechanism for signal geometries, enabling rate-distortion tradeoffs in which the distortion is measured on the signal geometry and not on the signal itself. In addition, the same embeddings can be used as a distributed coding mechanism, i.e., when information about the signal being coded is available in the decoder. Since embeddings can be used to represent only a limited amount of information, the same embedding design, when properly implemented, can also provide information-theoretic privacy guarantees. In summary, low-dimensional embeddings provide a fairly general framework for information representation, with wide applicability and relatively untapped potential.
Petros T. Boufounos is Senior Principal Research Scientist and the Computational Sensing Team Leader at Mitsubishi Electric Research Laboratories (MERL), and a visiting scholar at the Rice University Electrical and Computer Engineering department. Dr. Boufounos completed his undergraduate and graduate studies at MIT. He received the S.B. degree in Economics in 2000, the S.B. and M.Eng. degrees in Electrical Engineering and Computer Science (EECS) in 2002, and the Sc.D. degree in EECS in 2006. Between September 2006 and December 2008, he was a postdoctoral associate with the Digital Signal Processing Group at Rice University. Dr. Boufounos joined MERL in January 2009, where he has been heading the Computational Sensing Team since 2016.
Dr. Boufounos' immediate research focus includes signal acquisition and processing, inverse problems, frame theory, quantization and data representations. He is also interested into how signal acquisition interacts with other fields that use sensing extensively, such as machine learning, robotics and dynamical system theory. Dr. Boufounos has served as an Area Editor and a Senior Area Editor at IEEE Signal Processing Letters, has been part of the SigPort editorial board, and is currently a member of the IEEE Signal Processing Society Theory and Methods technical committee.
Host: Professor Donald Brown