Data science for networked data
Professor Po-Ling Loh, University of Wisconsin-Madison
We will survey a variety of problems involving mathematical analysis of network-structured data. In many scientific problems of contemporary interest, data are acquired in a very heterogeneous and non-i.i.d. fashion: Edges in a network may give rise to important correlations between node-level observations, which must be taken into account when performing data analysis. In large-scale applications, the structure of the graph may also determine the type of algorithms that may be performed. Our talk will cover topics such as influence maximization, source inference, graph hypothesis testing, immunization, and local optimization algorithms on networks.