Abstract: Inference of functions from data is ubiquitous in Signal Processing and Statistical Learning. This talk deals with Gaussian process (GP) based approaches that not only learn over a class of nonlinear functions, but also quantify the associated uncertainty. To cope with the curse of dimensionality, random feature Fourier (RF) vectors lead to parametric GP-RF function models, that offer scalable forms of Wiener’s minimum mean-square error approach. The talk will next touch upon deep GP architectures, and further focus on ensembles (E) of GP-RF learners, each with a distinct kernel belonging to a prescribed dictionary, and jointly learning a much richer class of functions. Whether in batch or online forms, EGPs remain robust to dynamics captured by adaptive Kalman filters. Their performance will be benchmarked using regret analysis. Broader applicability will be also demonstrated for policy evaluation in reinforcement learning. Case studies will highlight the merits of EGPs, and pay tribute to the ensemble of SP giants, namely Gauss, Fourier, Wiener, and Kalman.
Georgios B. Giannakis, Ph.D., is a University of Minnesota McKnight Presidential Chair in ECE, an Endowed Professor, and director of the Digital Technology Center. Giannakis has made fundamental contributions to statistical signal processing, wireless communications, networking, and data science, having major impact to several fields, including the smart power grid. He holds 33 US-patents; published 460 journal and 760 conference papers; 25 book chapters, two edited books and two research monographs, that have received 9 best journal paper awards, and more than 74,000 citations. Giannakis received the IEEE-SPS Nobert Wiener Award (2019); Technical Achievement Awards (SPS-2000, EURASIP-2005); the ComSoc Education Award (2019); and the IEEE Fourier Technical Field Award (2015). He is a Fellow of the IEEE, EURASIP, the US National Academy of Inventors, and has served the IEEE in several posts, including those of Editor-in-Chief, Board-of-Governors member, and Distinguished Lecturer.