Deterministic Anytime Inference for Continuous-Time Markov Processes
Date/Time: April 2, 3:40pm
Location: Lee Liu Hall Howe Hall
Continuous-time event data has become increasingly important in artificial intelligence and machine learning. Social networks, computer networks, phylogenetics, and server farm analytics all generate real-valued time stamps for events. In each of these areas, frequently we would like to either estimate an unseen portion of the events from the observations, or we would like to estimate the model of the system from partially-observed data streams.
In this talk, I will discuss a new method for exploiting time-ordered products to make such estimations both deterministic and anytime (convergent to the true answer in the limit of infinite computation time). This method combines the deterministic properties of variational approaches with the anytime properties of sampling. I will demonstrate results comparing it performance to the previous best methods available.
Dr. Shelton is an Associate Professor of Computer Science at the University of California at Riverside. He joined the faculty in 2003. He spent six months in 2003 and 2004 as a visiting faculty member at Intel Research and the 2012-13 academic year as a visiting researcher at Children's Hospital Los Angeles. He has been the Managing Editor of the Journal of Machine Learning Research and on the editorial board of the Editorial Board of the Journal of Artificial Intelligence Research.Christian R. Shelton received his B.S. in Computer Science from Stanford University in 1996. He then obtained his Ph.D. from MIT in 2001 and returned to Stanford from 2001 to 2003 as a post-doctoral scholar.Dr. Shelton's research interest is in statistical approaches to artificial intelligence, mainly in the areas of machine learning and dynamic processes. He also works at the intersection of learning and topics as varied as computer vision, sociology, game theory, decision theory, and computational biology.