Self-Organizing Map Based on Block Learning

Akitsugu OHTSUKA; Naotake KAMIURA; Teijiro ISOKAWA; Nobuyuki MATSUI

doi:10.1093/ietfec/e88-a.11.3151

Self-Organizing Map Based on Block Learning

Akitsugu OHTSUKA, Naotake KAMIURA, Teijiro ISOKAWA, Nobuyuki MATSUI

Full Text Views

0

Share
Cite this

Summary :

A block-matching-based self-organizing map (BMSOM) is presented. Finding a winner is carried out for each block, which is a set of neurons arranged in square. The proposed learning process updates the reference vectors of all of the neurons in a winner block. Then, the degrees of vector modifications are mainly controlled by the size (i.e., the number of neurons) of the winner block. To prevent a single cluster with neurons from splitting into some disjointed clusters, the restriction of the block size is imposed in the beginning of learning. At the main stage, this restriction is canceled. In BMSOM learning, the size of a winner block does not always decrease monotonically. The formula used to update the reference vectors is basically uncontrolled by time. Therefore, even if a map is in a nonstationary environment, training the map is probably pursued without interruption to adjust time-controlled parameters such as learning rate. Numerical results demonstrate that the BMSOM makes it possible to improve the plasticity of maps in a nonstationary environment and incremental learning.

Publication: IEICE TRANSACTIONS on Fundamentals Vol.E88-A No.11 pp.3151-3160

Publication Date: 2005/11/01

Publicized

Online ISSN

DOI: 10.1093/ietfec/e88-a.11.3151

Type of Manuscript: PAPER

Category: Nonlinear Problems

Cite this

Copy

Akitsugu OHTSUKA, Naotake KAMIURA, Teijiro ISOKAWA, Nobuyuki MATSUI, "Self-Organizing Map Based on Block Learning" in IEICE TRANSACTIONS on Fundamentals, vol. E88-A, no. 11, pp. 3151-3160, November 2005, doi: 10.1093/ietfec/e88-a.11.3151.
Abstract: A block-matching-based self-organizing map (BMSOM) is presented. Finding a winner is carried out for each block, which is a set of neurons arranged in square. The proposed learning process updates the reference vectors of all of the neurons in a winner block. Then, the degrees of vector modifications are mainly controlled by the size (i.e., the number of neurons) of the winner block. To prevent a single cluster with neurons from splitting into some disjointed clusters, the restriction of the block size is imposed in the beginning of learning. At the main stage, this restriction is canceled. In BMSOM learning, the size of a winner block does not always decrease monotonically. The formula used to update the reference vectors is basically uncontrolled by time. Therefore, even if a map is in a nonstationary environment, training the map is probably pursued without interruption to adjust time-controlled parameters such as learning rate. Numerical results demonstrate that the BMSOM makes it possible to improve the plasticity of maps in a nonstationary environment and incremental learning.
URL: https://globals.ieice.org/en_transactions/fundamentals/10.1093/ietfec/e88-a.11.3151/_p

Copy

@ARTICLE{e88-a_11_3151,
author={Akitsugu OHTSUKA, Naotake KAMIURA, Teijiro ISOKAWA, Nobuyuki MATSUI, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Self-Organizing Map Based on Block Learning},
year={2005},
volume={E88-A},
number={11},
pages={3151-3160},
abstract={A block-matching-based self-organizing map (BMSOM) is presented. Finding a winner is carried out for each block, which is a set of neurons arranged in square. The proposed learning process updates the reference vectors of all of the neurons in a winner block. Then, the degrees of vector modifications are mainly controlled by the size (i.e., the number of neurons) of the winner block. To prevent a single cluster with neurons from splitting into some disjointed clusters, the restriction of the block size is imposed in the beginning of learning. At the main stage, this restriction is canceled. In BMSOM learning, the size of a winner block does not always decrease monotonically. The formula used to update the reference vectors is basically uncontrolled by time. Therefore, even if a map is in a nonstationary environment, training the map is probably pursued without interruption to adjust time-controlled parameters such as learning rate. Numerical results demonstrate that the BMSOM makes it possible to improve the plasticity of maps in a nonstationary environment and incremental learning.},
keywords={},
doi={10.1093/ietfec/e88-a.11.3151},
ISSN={},
month={November},}

Copy

TY - JOUR
TI - Self-Organizing Map Based on Block Learning
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 3151
EP - 3160
AU - Akitsugu OHTSUKA
AU - Naotake KAMIURA
AU - Teijiro ISOKAWA
AU - Nobuyuki MATSUI
PY - 2005
DO - 10.1093/ietfec/e88-a.11.3151
JO - IEICE TRANSACTIONS on Fundamentals
SN -
VL - E88-A
IS - 11
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - November 2005
AB - A block-matching-based self-organizing map (BMSOM) is presented. Finding a winner is carried out for each block, which is a set of neurons arranged in square. The proposed learning process updates the reference vectors of all of the neurons in a winner block. Then, the degrees of vector modifications are mainly controlled by the size (i.e., the number of neurons) of the winner block. To prevent a single cluster with neurons from splitting into some disjointed clusters, the restriction of the block size is imposed in the beginning of learning. At the main stage, this restriction is canceled. In BMSOM learning, the size of a winner block does not always decrease monotonically. The formula used to update the reference vectors is basically uncontrolled by time. Therefore, even if a map is in a nonstationary environment, training the map is probably pursued without interruption to adjust time-controlled parameters such as learning rate. Numerical results demonstrate that the BMSOM makes it possible to improve the plasticity of maps in a nonstationary environment and incremental learning.
ER -