Almost-Smooth Nonparametric Regression and Pattern Recognition (3.2.2 Statistical Analysis and Methods; d. Bayesian and Non-parametric Statistics)
Abstract
Major Goals: Our project has two major thrusts: (1) We seek to establish methodology for nonparametric regression modeling with an almost smooth mean response function. This includes detection of jump discontinuities in the derivative of a mean response function of one variable (Subproject 1.1), in a mean response function of two variables (Subproject 1.2), and in a mean response function of two variables when a polar coordinate system can effectively reduce dimension (Subproject 1.3). (2) We seek to establish methodology for identifying the mean response function which describes the data generating mechanism in a nonparametric regression model. This includes study of convergence rates of misclassification probabilities (Subproject 2.1), examination of sampling schemes in this context (Subproject 2.2), and embedding the classification problem into an estimation problem via convex combinations of candidate functions (Subproject 2.3).
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 13, 2022
- Accession Number
- AD1195289
Entities
People
- Richard Charnigo
Organizations
- University of Kentucky