Abstract: Sampling is a standard approach to big graph analytics. But a good sample need to represent graph properties of interest with a known degree of accuracy. This talk describes a generic tunable sampling framework, graph sample and hold, that applies to graph stream sampling in which edges are presented one at a time, and from which unbiased estimators of graph properties can be produced in post-processing. The talk also describes the performance of the method on various types of graph, including social graphs, amongst others.
Speaker Bio: Nick Duffield is a Professor in the Department of Electrical and Computer Engineering at Texas A&M University. From 2013 until 2014, he was a Research Professor at DIMACS (the Center for Discrete Mathematics and Theoretical Computer Science) at Rutgers University, New Jersey, USA. From 1995 until 2013, he worked at AT&T Labs-Research where he was a Distinguished Member of Technical Staff and an AT&T Fellow.
Prof. Duffield works on the acquisition, analysis and applications of Big Data to communication networks and beyond.