Variable selection of high-dimensional non-parametric nonlinear systems: A way to avoid the curse of dimensionality

Er-wei Bai; Changmin Cheng; Wenxiao Zhao; Han-Fu Chen

doi:10.1109/CDC.2017.8264634

Back

Conference proceeding

Variable selection of high-dimensional non-parametric nonlinear systems: A way to avoid the curse of dimensionality

Er-wei Bai, Changmin Cheng, Wenxiao Zhao and Han-Fu Chen

2017 IEEE 56th Annual Conference on Decision and Control (CDC), Vol.2018-, pp.6469-6474

12/2017

DOI: 10.1109/CDC.2017.8264634

View Online

Abstract

This paper presents a numerically efficient algorithm for variable selection for high-dimensional nonlinear non-parametric systems. It is based on the average derivatives and relies on one dimensional estimates of the density function and its derivative. Thus it avoids the curse of dimensionality usually encountered for high-dimensional systems. Theoretical analysis is provided and shows that the conditions derived are sufficient for a variable to contribute, and necessary and almost sufficient for a variable not to contribute for a class of nonlinear systems. Further, convergence results are established.

Optimization

Convergence

Density functional theory

Input variables

Kernel

Nonlinear systems

Reliability

Details

Title: Subtitle: Variable selection of high-dimensional non-parametric nonlinear systems: A way to avoid the curse of dimensionality
Creators: Er-wei Bai - University of Iowa
Changmin Cheng - University of Iowa
Wenxiao Zhao - University of Iowa
Han-Fu Chen - University of Iowa
Resource Type: Conference proceeding
Publication Details: 2017 IEEE 56th Annual Conference on Decision and Control (CDC), Vol.2018-, pp.6469-6474
DOI: 10.1109/CDC.2017.8264634
Publisher: IEEE
Language: English
Date published: 12/2017
Academic Unit: Electrical and Computer Engineering
Record Identifier: 9984197194602771

Metrics

16 Record Views