A Random Matrix Framework for BigData Machine Learning, and Applications to Wireless Communications

Romain Couillet - Full Professor in the LSS laboratory at CentraleSupélec, France
Communication systems

Date: June 14th 2017
Location: Eurecom - Eurecom

Abstract: The recent interest for automating classification, clustering, detection, estimation of the massively large and heterogeneous data made available by modern computer and communication technologies, and the fast deployment of deep learning approaches, has spurred a new need for mathematical tools to apprehend bigdata machine learning. In this talk, I will argue that random matrix theory (RMT) is a key enabler to meet this challenge. After a brief introduction of the basic concepts, I will show how RMT helps understand and improve spectral clustering methods (notably kernel data clustering), semi-supervised learning and neural networks alike. Beyond the theoretical considerations, I shall exemply the actual machine learning performance gains achieved by the RMT framework through examples in image classification, graph labeling, but also specifically in massive MIMO wireless communications. Related references : (available at http://romaincouillet.hebfree.org/publications.html) * C. Louart, Z. Liao, R. Couillet, "A Random Matrix Approach to Neural Networks", (to appear in) Annals of Applied Probability, 2017. * R. Couillet, F. Benaych-Georges, "Kernel Spectral Clustering of Large Dimensional Data", Electronic Journal of Statistics, vol. 10, no. 1, pp. 1393-1454, 2016. * X. Mai, R. Couillet, "The counterintuitive mechanism of graph-based semi-supervised learning in the big data regime", IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP'17), New Orleans, USA, 2017. * R. Couillet, A. Kammoun, "Random Matrix Improved Subspace Clustering", Asilomar Conference on Signals, Systems, and Computers, Pacific Grove, CA, USA, 2016. * Z. Liao, R. Couillet, "A Large Dimensional Analysis of Least Squares Support Vector Machines", (under revision) Journal of Machine Learning Research, 2017. * R. Couillet, M. Debbah, "Random matrix theory methods for wireless communications", Cambridge University Press, 2011. Short bio : Romain Couillet received his MSc in Mobile Communications at the Eurecom Institute and his MSc in Communication Systems in Telecom ParisTech, France in 2007. From 2007 to 2010, he worked with ST-Ericsson as an Algorithm Development Engineer on the Long Term Evolution Advanced project, where he prepared his PhD with Supelec, France, which he graduated in November 2010. He is currently a full professor in the LSS laboratory at CentraleSupélec, France. His research topics are in random matrix theory applied to statistics, machine learning, signal processing, and wireless communications. He is an IEEE Senior Member. In 2015, he received the HDR title from University ParisSud. He is the recipient of the 2013 CNRS Bronze Medal in the section "science of information and its interactions", of the 2013 IEEE ComSoc Outstanding Young Researcher Award (EMEA Region), of the 2011 EEA/GdR ISIS/GRETSI best PhD thesis award, and of the Valuetools 2008 best student paper award.

Permalink: https://www.eurecom/seminar/62000