A0472
Title: Nonparametric logistic regression in high dimension
Authors: Jong Soo Lee - University of Massachusetts Lowell (United States) [presenting]
Johan Lim - Seoul National University (Korea, South)
Abstract: The maximum penalized likelihood estimation for logistic regression in high dimension is studied. When dealing with high dimensional regression problem, one uses either the dimension reduction techniques in input variables or use penalized approach using ridge, LASSO, or elastic net as penalty functions. We propose a new penalty function for logistic regression that alleviates some of the issues in ridge or LASSO type penalties, but at the same time preserves the optimal convergence rate under mild conditions.
