Date of Award

Summer 2006

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Electrical & Computer Engineering

Program/Concentration

Electrical Engineering

Committee Director

Stephen A. Zahorian

Committee Member

Vijayan K. Asari

Committee Member

Min Song

Call Number for Print

Special Collections LD4331.E55 S57 2006

Abstract

Advances in data collection and storage capabilities during the past decades have led to an information overload in most sciences. Traditional statistical methods break down partly because of the increase in the number of observations, but mostly because of the increase in the number of variables associated with each observation. While certain methods can construct predictive models with high accuracy from high-dimensional data, it is still of interest in many applications to reduce the dimension of the original data prior to any modeling of the data. Patterns in the data can be hard to find in data of high dimensionality, where the luxury of graphical representation is not available. Linear PCA is a powerful tool for analyzing this high-dimensional data. A common drawback of these classical methods is that only linear structures can be correctly extracted from the data.

If the data represent the complicated interaction of features, then a linear subspace may be a poor representation and a nonlinear subspace may be needed. It is hypothesized that nonlinear ( curved) basis vectors will be more efficient for representing some types of data than are linear PCA basis vectors. The neural network method is among the best methods to implement NLPCA due to its capability of accurately approximating any continuous nonlinear function. Linear and nonlinear PCA are compared using speech classification experiments with reduced dimensionality data. For some cases, NLPCA is found to be advantageous over classical PCA.

Rights

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

DOI

10.25777/7b1p-6m64

Download

Share

COinS