Events Calendar

Network Approaches to Data-Driven Problems: Fundamental Limits, Scalable Algorithms, and Applications

Speaker: Soheil Feizi, Massachusetts Institute of Technology

Talk Title: Network Approaches to Data-Driven Problems: Fundamental Limits, Scalable Algorithms, and Applications

Abstract: In large-scale data-driven problems, network modeling provides a unifying framework to succinctly represent data, reveal underlying data structures, and facilitate experiment design. In practice, however, size, uncertainty and complexity of the underlying associations render these applications challenging. In this talk, I will illustrate the use of spectral, combinatorial, and statistical inference techniques in learning the network topology and subsequent network analysis.

First, we introduce Network Maximal Correlation (NMC), a multivariate measure of nonlinear association suitable for large datasets. NMC infers transformations of variables to reveal underlying nonlinear dependencies among them. We characterize NMC using geometric properties of Hilbert spaces and illustrate its application in learning graphical models when variables have unknown nonlinear dependencies. Next, we discuss the problem of network alignment that aims to find a bijective mapping across two graphs so that, if two nodes are connected in one graph, their images are also connected in the other graph. This problem has a broad range of applications for comparative network analysis in systems biology, social sciences and engineering areas. To solve this combinatorial problem, we present a new scalable spectral algorithm, and establish its efficiency, theoretically and experimentally, over several synthetic and real networks.

Biography: Soheil Feizi is a Ph.D. candidate in the Electrical Engineering and Computer Science (EECS) Department at Massachusetts Institute of Technology (MIT), co-supervised by Prof. Muriel Médard and Prof. Manolis Kellis. His research focuses on complex network analysis using tools and concepts from optimization, machine learning, statistical inference and information theory. Previously, he completed a M.Sc. in Electrical Engineering at MIT, where he received the Jacobs Presidential Fellowship and EECS Great Educators Fellowship, as well as an Ernst Guillemin Award for his Master of Science Thesis. He also received the best student award in Electrical Engineering at Sharif University of Technology from where he holds his B.Sc.

Host: Salman Avestimehr

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

Audiences: Everyone Is Invited

Contact: Suzanne Wong