The University of Chicago

Abstract:
The graphical model has proven to be a useful abstraction in statistics and machine learning. The starting point is the graph of a distribution. While often the graph is assumed given, we have been studying the problem of estimating the graph from data. In this talk we present several nonparametric and semiparametric methods for graph estimation. One approach is a nonparametric extension of the Gaussian graphical model that allows arbitrary graphs. For the discrete Gaussian (Ising model), we use parallel neighborhood selection with L1-regularized logistic regression. Alternatively, we can restrict the family of graphs to spanning forests, enabling the use of fully nonparametric density estimation in high dimensions. When additional covariates are available, we propose a framework for graph-valued regression. The resulting methods are easy to understand and use, theoretically well supported, and effective for modeling and exploring high dimensional data. Joint work with Han Liu, Pradeep Ravikumar, Martin Wainwright, and Larry Wasserman.

Bio:
John Lafferty is the Louis Block Professor in the Departments of Statistics, Computer Science, and the College at The University of Chicago. His research area is machine learning, with a focus on computational and statistical aspects of nonparametric methods, high-dimensional data, graphical models, and applications. An associate editor of the Journal of Machine Learning Research, Dr. Lafferty served as progam co-chair and general co-chair of the Neural Information Processing Systems Foundation conferences in 2009 and 2010. Dr. Lafferty received his doctoral degree in mathematics from Princeton University, where he was a member of the Program in Applied and Computational Mathematics. Prior to joining the University of Chicago in 2011, he was Professor of Computer Science, Machine Learning, and Statistics at Carnegie Mellon University, where he is currently an Adjunct Professor.