Guest Speaker: Purnamrita Sarkar
University of California, Berkeley
Talk Abstract: Many real world networks change over time, and for a variety of network-based applications it is important to predict future interaction between a pair of entities. In this talk I will discuss computationally efficient algorithms for nearest neighbor search in large graphs in the context of link prediction. I will also provide theoretical insight for the usefulness of popular link prediction heuristics. If time permits, I will discuss generative models we have developed for dynamically evolving networks.
About the Speaker: Purnamrita Sarkar is currently a postdoctoral scholar at U. C. Berkeley jointly in the Department of Electrical Engineering and Computer Science and the Department of Statistics. She received her Bachelor's degree in 2004 from the Computer Science and Engineering department, Indian Institute of Technology, Kharagpur, and her PhD degree from the Machine Learning Department, at Carnegie Mellon University in 2010.
Host: Yusu Wang
* Purnamrita Sarkar is a CSE faculty candidate