Selection of the optimal parameter value for the Isomap algorithm

被引：101

作者：

Samko, O. ^{[1
]}

Marshall, A. D. ^{[1
]}

Rosin, P. L. ^{[1
]}

机构：

[1] Cardiff Univ, Sch Comp Sci, Cardiff CF24 3AA, Wales

来源：

PATTERN RECOGNITION LETTERS | 2006年 / 27卷 / 09期

关键词：

nonlinear dimensionality reduction; manifold learning; Isomap;

D O I：

10.1016/j.patrec.2005.11.017

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

The isometric feature mapping (Isomap) method has demonstrated promising results in finding low-dimensional manifolds from data points in high-dimensional input space. Isomap has one free parameter (number of nearest neighbours K or neighbourhood radius 6), which has to be specified manually. In this paper we present a new method for selecting the optimal parameter value for Isomap automatically. Numerous experiments on synthetic and real data sets show the effectiveness of our method. (c) 2006 Elsevier B.V. All rights reserved.

引用

页码：968 / 979

页数：12

共 22 条

[1] [Anonymous], 2003, EUR S ART NEUR NETW
[2] [Anonymous], 1952, Psychometrika
[3] Laplacian eigenmaps for dimensionality reduction and data representation
Belkin, M
Niyogi, P
[J]. NEURAL COMPUTATION, 2003, 15 (06) : 1373 - 1396
[4] BRAND M, 2004, TR2004134 MERL RES L
[5] Estimating the intrinsic dimension of data with a fractal-based method
Camastra, F
Vinciarelli, A
[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2002, 24 (10) : 1404 - 1407
[6] Geodesic entropic graphs for dimension and entropy estimation in manifold learning
Costa, JA
Hero, AO
[J]. IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2004, 52 (08) : 2210 - 2221
[7] de Ridder D, 2003, LECT NOTES COMPUT SC, V2714, P333
[8] De Ridder D., 2002, PH200201 DELFT U TEC, P1
[9] Hessian eigenmaps: Locally linear embedding techniques for high-dimensional data
Donoho, DL
Grimes, C
[J]. PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2003, 100 (10) : 5591 - 5596
[10] FRIEDRICH T, 2002, THESIS U SHEFFIELD

← 1 2 3 →