Finding the number of clusters for nonparametric segmentation

N Nasios; A G Bors

Finding the number of clusters for nonparametric segmentation

Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Non-parametric data representation can be done by means of a potential function. This paper introduces a methodology for finding modes of the potential function. Two different methods are considered for the potential function representation: by using summations of Gaussian kernels, and by employing quantum clustering. In the second case each data sample is associated with a quantum physics particle that has a radial energy field around its location. Both methods use a scaling parameter (bandwidth) to model the strength of the influence around each data sample. We estimate the scaling parameter as the mean of the Gamma distribution that models the variances of K-nearest data samples to any given data. The local Hessian is used afterwards to find the modes of the resulting potential function. Each mode is associated with a cluster. We apply the proposed algorithm for blind signal separation and for the topographic segmentation of radar images of terrain.

Original language	English
Title of host publication	COMPUTER ANALYSIS OF IMAGES AND PATTERNS, PROCEEDINGS
Editors	A Gagalowicz, W Philips
Place of Publication	BERLIN
Publisher	Springer
Pages	213-221
Number of pages	9
ISBN (Print)	3-540-28969-0
Publication status	Published - 2005
Event	11th International Conference on Computer Analysis of Images and Patterns - Versailles Duration: 5 Sept 2005 → 8 Sept 2005

Conference

Conference	11th International Conference on Computer Analysis of Images and Patterns
City	Versailles
Period	5/09/05 → 8/09/05

Cite this

@inproceedings{20af72fca91642268ccaf997834cd299,

title = "Finding the number of clusters for nonparametric segmentation",

abstract = "Non-parametric data representation can be done by means of a potential function. This paper introduces a methodology for finding modes of the potential function. Two different methods are considered for the potential function representation: by using summations of Gaussian kernels, and by employing quantum clustering. In the second case each data sample is associated with a quantum physics particle that has a radial energy field around its location. Both methods use a scaling parameter (bandwidth) to model the strength of the influence around each data sample. We estimate the scaling parameter as the mean of the Gamma distribution that models the variances of K-nearest data samples to any given data. The local Hessian is used afterwards to find the modes of the resulting potential function. Each mode is associated with a cluster. We apply the proposed algorithm for blind signal separation and for the topographic segmentation of radar images of terrain.",

author = "N Nasios and Bors, {A G}",

year = "2005",

language = "English",

isbn = "3-540-28969-0",

pages = "213--221",

editor = "A Gagalowicz and W Philips",

booktitle = "COMPUTER ANALYSIS OF IMAGES AND PATTERNS, PROCEEDINGS",

publisher = "Springer",

address = "Germany",

note = "11th International Conference on Computer Analysis of Images and Patterns ; Conference date: 05-09-2005 Through 08-09-2005",

}

TY - GEN

T1 - Finding the number of clusters for nonparametric segmentation

AU - Nasios, N

AU - Bors, A G

PY - 2005

Y1 - 2005

N2 - Non-parametric data representation can be done by means of a potential function. This paper introduces a methodology for finding modes of the potential function. Two different methods are considered for the potential function representation: by using summations of Gaussian kernels, and by employing quantum clustering. In the second case each data sample is associated with a quantum physics particle that has a radial energy field around its location. Both methods use a scaling parameter (bandwidth) to model the strength of the influence around each data sample. We estimate the scaling parameter as the mean of the Gamma distribution that models the variances of K-nearest data samples to any given data. The local Hessian is used afterwards to find the modes of the resulting potential function. Each mode is associated with a cluster. We apply the proposed algorithm for blind signal separation and for the topographic segmentation of radar images of terrain.

AB - Non-parametric data representation can be done by means of a potential function. This paper introduces a methodology for finding modes of the potential function. Two different methods are considered for the potential function representation: by using summations of Gaussian kernels, and by employing quantum clustering. In the second case each data sample is associated with a quantum physics particle that has a radial energy field around its location. Both methods use a scaling parameter (bandwidth) to model the strength of the influence around each data sample. We estimate the scaling parameter as the mean of the Gamma distribution that models the variances of K-nearest data samples to any given data. The local Hessian is used afterwards to find the modes of the resulting potential function. Each mode is associated with a cluster. We apply the proposed algorithm for blind signal separation and for the topographic segmentation of radar images of terrain.

M3 - Conference contribution

SN - 3-540-28969-0

SP - 213

EP - 221

BT - COMPUTER ANALYSIS OF IMAGES AND PATTERNS, PROCEEDINGS

A2 - Gagalowicz, A

A2 - Philips, W

PB - Springer

CY - BERLIN

T2 - 11th International Conference on Computer Analysis of Images and Patterns

Y2 - 5 September 2005 through 8 September 2005

ER -