Online Program Home
My Program

Abstract Details

Activity Number: 254 - Contributed Poster Presentations: Section on Statistical Learning and Data Science
Type: Contributed
Date/Time: Monday, July 30, 2018 : 2:00 PM to 3:50 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #329748
Title: Sparse Variable Selection in Kernel Discriminant Analysis via Optimal Scoring
Author(s): Alexander Lapanowski* and Irina Gaynanova
Companies: Texas A&M and Texas A&M Univeristy
Keywords: Reproducing kernel Hilbert Space; Variable selection; Kernel classification; Kernel ridge regression; LASSO

We consider the two-group classification problem and propose a new nonparametric algorithm for simultaneous variable selection and nonlinear dimension reduction. Our approach is based on applying the kernel trick within an optimal scoring framework and imposing structured sparsity using weighted kernels. The use of an optimal scoring framework enables efficient computations, and sparsity allows the removal of noisy features from the classification rule. Numerical studies demonstrate the superior classification performance of the proposed approach compared to common nonparametric classifiers. We support our empirical findings with theoretical guarantees on the expected risk consistency of the method.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program