Online Program Home
  My Program

Abstract Details

Activity Number: 669 - Recent Advances in Nonparametric Statistics
Type: Contributed
Date/Time: Thursday, August 3, 2017 : 10:30 AM to 12:20 PM
Sponsor: Section on Nonparametric Statistics
Abstract #323471
Author(s): Richard Charnigo* and Cidambi Srinivasan
Companies: University of Kentucky and University of Kentucky
Keywords: nonparametric regression ; classification ; pattern recognition ; characterization ; nanoparticle ; model selection

Given values of a covariate X, suppose we observe values of a response Y from one of several nonparametric data-generating regimes: a mean response function mu_1 plus noise, another mean response function mu_2 plus noise, and so forth. Suppose that mu_1, mu_2, etc., are known but that we are not certain which one is generating the values of Y. Two questions arise. First, how can we infer the data-generating mechanism? Second, if we can choose the values of X, how shall we do so? One possible approach is to space the values of X along a grid, then use the values of Y to nonparametrically estimate the data-generating mechanism, and finally compare the estimated data-generating regime to mu_1, mu_2, etc. In this work, however, we show that the data-generating mechanism can be inferred without nonparametric estimation, such that the risk of misclassification decays at an exponential rate with respect to the sample size. We discuss the implications for addressing an inverse problem such as ascertaining nanoparticle properties from scattering data.

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

Back to the full JSM 2017 program

Copyright © American Statistical Association