The primary goal of communication is to exchange information between machines rapidly and reliably. Increasing the amount of data in recent years has posed a significant challenge on reliable communication. However, in many cases, the knowledge embedded in the data is what the receiver is seeking for not the data itself. An instance of this problem is learning of tree-structured Gaussian graphical models from data distributed across a network. We will explore how by sending few bits per sample to a central machine will enable us to discover the underlying structure accurately. Simulation results on both synthetic and real-world datasets show that our strategies achieve a desired accuracy in inferring the underlying structure.
Seyed Abolfazl Motahari received the B.Sc. degreefrom Iran University of Science and Technology (IUST), Tehran, Iran, in 1999, the M.Sc. degree fromSharif University of Technology (SUT), Tehran, Iran, in 2001, and the Ph.D. degree from University of Waterloo, Waterloo, ON, Canada, in 2009,all in electrical engineering. He is currently an Assistant Professor with the Department of Computer Engineering, SUT. From October 2009 to September 2010, he was a Postdoctoral Fellow with the University of Waterloo. From September 2010 to July 2013, he was a Postdoctoral Fellow with the Department of Electrical Engineering and Computer Sciences, University of California at Berkeley. His research interests include multiuser information theory and bioinformatics. He received severalawards including the Natural Science and Engineering Research Council ofCanada Postdoctoral Fellowship.