You are here

ANALYSIS OF KOLMOGOROV'S SUPERPOSITION THEOREM AND ITS IMPLEMENTATION IN APPLICATIONS WITH LOW AND HIGH DIMENSIONAL DATA.

Download pdf | Full Screen View

Date Issued:
2008
Abstract/Description:
In this dissertation, we analyze Kolmogorov's superposition theorem for high dimensions. Our main goal is to explore and demonstrate the feasibility of an accurate implementation of Kolmogorov's theorem. First, based on Lorentz's ideas, we provide a thorough discussion on the proof and its numerical implementation of the theorem in dimension two. We present computational experiments which prove the feasibility of the theorem in applications of low dimensions (namely, dimensions two and three). Next, we present high dimensional extensions with complete and detailed proofs and provide the implementation that aims at applications with high dimensionality. The amalgamation of these ideas is evidenced by applications in image (two dimensional) and video (three dimensional) representations, the content based image retrieval, video retrieval, de-noising and in-painting, and Bayesian prior estimation of high dimensional data from the fields of computer vision and image processing.
Title: ANALYSIS OF KOLMOGOROV'S SUPERPOSITION THEOREM AND ITS IMPLEMENTATION IN APPLICATIONS WITH LOW AND HIGH DIMENSIONAL DATA.
70 views
45 downloads
Name(s): Bryant, Donald, Author
Li, Xin, Committee Chair
University of Central Florida, Degree Grantor
Type of Resource: text
Date Issued: 2008
Publisher: University of Central Florida
Language(s): English
Abstract/Description: In this dissertation, we analyze Kolmogorov's superposition theorem for high dimensions. Our main goal is to explore and demonstrate the feasibility of an accurate implementation of Kolmogorov's theorem. First, based on Lorentz's ideas, we provide a thorough discussion on the proof and its numerical implementation of the theorem in dimension two. We present computational experiments which prove the feasibility of the theorem in applications of low dimensions (namely, dimensions two and three). Next, we present high dimensional extensions with complete and detailed proofs and provide the implementation that aims at applications with high dimensionality. The amalgamation of these ideas is evidenced by applications in image (two dimensional) and video (three dimensional) representations, the content based image retrieval, video retrieval, de-noising and in-painting, and Bayesian prior estimation of high dimensional data from the fields of computer vision and image processing.
Identifier: CFE0002236 (IID), ucf:47909 (fedora)
Note(s): 2008-08-01
Ph.D.
Sciences, Department of Mathematics
Doctorate
This record was generated from author submitted information.
Subject(s): Kolmogorov
superposition
image
processing
composition
Persistent Link to This Record: http://purl.flvc.org/ucf/fd/CFE0002236
Restrictions on Access: public
Host Institution: UCF

In Collections