Kohonen Learning with a Mechanism, the Law of the Jungle, Capable of Dealing with Nonstationary Probability Distribution Functions

Taira NAKAJIMA; Hiroyuki TAKIZAWA; Hiroaki KOBAYASHI; Tadao NAKAMURA

Kohonen Learning with a Mechanism, the Law of the Jungle, Capable of Dealing with Nonstationary Probability Distribution Functions

Taira NAKAJIMA, Hiroyuki TAKIZAWA, Hiroaki KOBAYASHI, Tadao NAKAMURA

Full Text Views

0

Share
Cite this

Summary :

We present a mechanism, named the law of the jungle (LOJ), to improve the Kohonen learning. The LOJ is used to be an adaptive vector quantizer for approximating nonstationary probability distribution functions. In the LOJ mechanism, the probability that each node wins in a competition is dynamically estimated during the learning. By using the estimated win probability, "strong" nodes are increased through creating new nodes near the nodes, and "weak" nodes are decreased through deleting themselves. A pair of creation and deletion is treated as an atomic operation. Therefore, the nodes which cannot win the competition are transferred directly from the region where inputs almost never occur to the region where inputs often occur. This direct "jump" of weak nodes provides rapid convergence. Moreover, the LOJ requires neither time-decaying parameters nor a special periodic adaptation. From the above reasons, the LOJ is suitable for quick approximation of nonstationary probability distribution functions. In comparison with some other Kohonen learning networks through experiments, only the LOJ can follow nonstationary probability distributions except for under high-noise environments.

Publication: IEICE TRANSACTIONS on Information Vol.E81-D No.6 pp.584-591

Publication Date: 1998/06/25

Publicized

Online ISSN

DOI

Type of Manuscript

Category: Bio-Cybernetics and Neurocomputing

Cite this

Copy

Taira NAKAJIMA, Hiroyuki TAKIZAWA, Hiroaki KOBAYASHI, Tadao NAKAMURA, "Kohonen Learning with a Mechanism, the Law of the Jungle, Capable of Dealing with Nonstationary Probability Distribution Functions" in IEICE TRANSACTIONS on Information, vol. E81-D, no. 6, pp. 584-591, June 1998, doi: .
Abstract: We present a mechanism, named the law of the jungle (LOJ), to improve the Kohonen learning. The LOJ is used to be an adaptive vector quantizer for approximating nonstationary probability distribution functions. In the LOJ mechanism, the probability that each node wins in a competition is dynamically estimated during the learning. By using the estimated win probability, "strong" nodes are increased through creating new nodes near the nodes, and "weak" nodes are decreased through deleting themselves. A pair of creation and deletion is treated as an atomic operation. Therefore, the nodes which cannot win the competition are transferred directly from the region where inputs almost never occur to the region where inputs often occur. This direct "jump" of weak nodes provides rapid convergence. Moreover, the LOJ requires neither time-decaying parameters nor a special periodic adaptation. From the above reasons, the LOJ is suitable for quick approximation of nonstationary probability distribution functions. In comparison with some other Kohonen learning networks through experiments, only the LOJ can follow nonstationary probability distributions except for under high-noise environments.
URL: https://globals.ieice.org/en_transactions/information/10.1587/e81-d_6_584/_p

Copy

@ARTICLE{e81-d_6_584,
author={Taira NAKAJIMA, Hiroyuki TAKIZAWA, Hiroaki KOBAYASHI, Tadao NAKAMURA, },
journal={IEICE TRANSACTIONS on Information},
title={Kohonen Learning with a Mechanism, the Law of the Jungle, Capable of Dealing with Nonstationary Probability Distribution Functions},
year={1998},
volume={E81-D},
number={6},
pages={584-591},
abstract={We present a mechanism, named the law of the jungle (LOJ), to improve the Kohonen learning. The LOJ is used to be an adaptive vector quantizer for approximating nonstationary probability distribution functions. In the LOJ mechanism, the probability that each node wins in a competition is dynamically estimated during the learning. By using the estimated win probability, "strong" nodes are increased through creating new nodes near the nodes, and "weak" nodes are decreased through deleting themselves. A pair of creation and deletion is treated as an atomic operation. Therefore, the nodes which cannot win the competition are transferred directly from the region where inputs almost never occur to the region where inputs often occur. This direct "jump" of weak nodes provides rapid convergence. Moreover, the LOJ requires neither time-decaying parameters nor a special periodic adaptation. From the above reasons, the LOJ is suitable for quick approximation of nonstationary probability distribution functions. In comparison with some other Kohonen learning networks through experiments, only the LOJ can follow nonstationary probability distributions except for under high-noise environments.},
keywords={},
doi={},
ISSN={},
month={June},}

Copy

TY - JOUR
TI - Kohonen Learning with a Mechanism, the Law of the Jungle, Capable of Dealing with Nonstationary Probability Distribution Functions
T2 - IEICE TRANSACTIONS on Information
SP - 584
EP - 591
AU - Taira NAKAJIMA
AU - Hiroyuki TAKIZAWA
AU - Hiroaki KOBAYASHI
AU - Tadao NAKAMURA
PY - 1998
DO -
JO - IEICE TRANSACTIONS on Information
SN -
VL - E81-D
IS - 6
JA - IEICE TRANSACTIONS on Information
Y1 - June 1998
AB - We present a mechanism, named the law of the jungle (LOJ), to improve the Kohonen learning. The LOJ is used to be an adaptive vector quantizer for approximating nonstationary probability distribution functions. In the LOJ mechanism, the probability that each node wins in a competition is dynamically estimated during the learning. By using the estimated win probability, "strong" nodes are increased through creating new nodes near the nodes, and "weak" nodes are decreased through deleting themselves. A pair of creation and deletion is treated as an atomic operation. Therefore, the nodes which cannot win the competition are transferred directly from the region where inputs almost never occur to the region where inputs often occur. This direct "jump" of weak nodes provides rapid convergence. Moreover, the LOJ requires neither time-decaying parameters nor a special periodic adaptation. From the above reasons, the LOJ is suitable for quick approximation of nonstationary probability distribution functions. In comparison with some other Kohonen learning networks through experiments, only the LOJ can follow nonstationary probability distributions except for under high-noise environments.
ER -