Princeton University

School of Engineering & Applied Science

Minimum Distance Estimation for Robust and Sparse High-Dimensional Regression

Speaker: 
Aurélie Lozano, IBM T.J. Watson Research Center
Location: 
Engineering Quadrangle B205
Date/Time: 
Tuesday, November 19, 2013 - 4:30pm

Abstract: We propose a minimum distance estimation method for robust regression in sparse high-dimensional settings. The traditional likelihood-based estimators lack resilience against outliers, a critical issue when dealing with high-dimensional noisy data. Our method, Minimum Distance Lasso (MD-Lasso), combines minimum distance functionals, customarily used in nonparametric estimation for their robustness, with l1-regularization for high-dimensional regression. The geometry of MD-Lasso is key to its consistency and robustness. The estimator is governed by a scaling parameter that caps the influence of outliers: the loss per observation is locally convex and close to quadratic for small squared residuals, and flattens for squared residuals larger than the scaling parameter. As the parameter approaches infinity, the estimator becomes equivalent to least-squares Lasso. MD-Lasso enjoys fast convergence rates under mild conditions on the model error distribution, which hold for any of the solutions in a convexity region around the true parameter and in certain cases for every solution. Remarkably, a first-order optimization method is able to produce iterates very close to the consistent solutions, with geometric convergence and regardless of the initialization. A connection is established with re-weighted least-squares that intuitively explains MD-Lasso robustness. The merits of our method are demonstrated through simulation and eQTL data analysis.

Joint work with Nicolai Meinshausen, Professor of Statistics, University of Oxford and ETH Zürich.

Biography: Aurélie Lozano received her Ph.D. from Princeton University, where she was a recipient of the Gordon Y.S. Wu Fellowship. Since 2007, she has been a research staff member in the Machine Learning group at the IBM T.J. Watson Research Center, Yorktown Heights, NY. Dr. Lozano's research interests include machine learning, statistics and data mining. Her current focus is on methods for solving high dimensional data problems, their theoretical analysis, and applications to computational biology, environmental sciences, and social media analytics.