The rise of social and Internet networks with hundreds of millions to billions of nodes, presents new challenges for scaling up statistical network models to execute in a reasonable amount of time on Internet-scale networks. By applying a succinct representation of networks as a bag of triangular motifs, developing a parsimonious statistical model, deriving an efficient stochastic variational inference algorithm, and implementing it as a distributed cluster program, we demonstrate latent space inference and overlapping community detection on very large networks with over 100 million nodes on just a few cluster machines. Compared to other state-of-the-art probabilistic network approaches, our method is several orders of magnitude faster, with competitive or improved accuracy at overlapping community detection. This is joint work with Qirong Ho and Eric P. Xing.
Junming Yin is an assistant professor in the Eller College of Management at University of Arizona. Prior to that he was a Lane Fellow in the School of Computer Science at Carnegie Mellon University. He received his Ph.D. in Computer Science and M.A. in Statistics from UC Berkeley under the advising of Prof. Michael I. Jordan and Prof. Yun S. Song. His research interests focus on statistical machine learning and its applications in business intelligence, digital marketing, biology and healthcare.