A weighted grey CMAC neural network with output differentiability

Chih Ming Chen, Chin Ming Hong

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

The Cerebellar Model Arithmetic Computer (CMAC) is a table lookup neurocomputing technique. It can be viewed as a basis function network (BFN) and performs well in terms of its fast learning speed, local generalization capability for approximating nonlinear functions. However, a disvantage is that the derivative of its output cannot be preserved due to the CMAC uses a constant basis function within each quantized state. This creates the limitation and inconvenience while the derivative information is needed in real-world applications. This paper proposes a Weight Grey CMAC (WGCMAC) that includes the conventional CMAC weight addressing scheme and the weighted grey prediction model to resolve this problem. Based on the weighted grey prediction model, we present an efficient learning algorithm for the proposed WGCMAC. Experiments confirm that the WGCMAC not only has a faster learning speed than the conventional CMAC, but also provides output derivatives and more precise learning result. Besides, compared with other enhanced CMAC models providing output derivatives, the proposed method has fastest learning speed.

Original languageEnglish
Title of host publicationAnnual Conference of the North American Fuzzy Information Processing Society - NAFIPS
EditorsM.H. Smith, W.A. Gruver, L.O. Hall
Pages1009-1014
Number of pages6
Volume2
Publication statusPublished - 2001
EventJoint 9th IFSA World Congress and 20th NAFIPS International Conference - Vancouver, BC, Canada
Duration: 2001 Jul 252001 Jul 28

Other

OtherJoint 9th IFSA World Congress and 20th NAFIPS International Conference
CountryCanada
CityVancouver, BC
Period01/7/2501/7/28

Fingerprint

Neural networks
Derivatives
Table lookup
Learning algorithms
Experiments

ASJC Scopus subject areas

  • Computer Science(all)
  • Media Technology

Cite this

Chen, C. M., & Hong, C. M. (2001). A weighted grey CMAC neural network with output differentiability. In M. H. Smith, W. A. Gruver, & L. O. Hall (Eds.), Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS (Vol. 2, pp. 1009-1014)

A weighted grey CMAC neural network with output differentiability. / Chen, Chih Ming; Hong, Chin Ming.

Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS. ed. / M.H. Smith; W.A. Gruver; L.O. Hall. Vol. 2 2001. p. 1009-1014.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Chen, CM & Hong, CM 2001, A weighted grey CMAC neural network with output differentiability. in MH Smith, WA Gruver & LO Hall (eds), Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS. vol. 2, pp. 1009-1014, Joint 9th IFSA World Congress and 20th NAFIPS International Conference, Vancouver, BC, Canada, 01/7/25.
Chen CM, Hong CM. A weighted grey CMAC neural network with output differentiability. In Smith MH, Gruver WA, Hall LO, editors, Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS. Vol. 2. 2001. p. 1009-1014
Chen, Chih Ming ; Hong, Chin Ming. / A weighted grey CMAC neural network with output differentiability. Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS. editor / M.H. Smith ; W.A. Gruver ; L.O. Hall. Vol. 2 2001. pp. 1009-1014
@inproceedings{7bcb53903dca4348bc644d2721c9b37c,
title = "A weighted grey CMAC neural network with output differentiability",
abstract = "The Cerebellar Model Arithmetic Computer (CMAC) is a table lookup neurocomputing technique. It can be viewed as a basis function network (BFN) and performs well in terms of its fast learning speed, local generalization capability for approximating nonlinear functions. However, a disvantage is that the derivative of its output cannot be preserved due to the CMAC uses a constant basis function within each quantized state. This creates the limitation and inconvenience while the derivative information is needed in real-world applications. This paper proposes a Weight Grey CMAC (WGCMAC) that includes the conventional CMAC weight addressing scheme and the weighted grey prediction model to resolve this problem. Based on the weighted grey prediction model, we present an efficient learning algorithm for the proposed WGCMAC. Experiments confirm that the WGCMAC not only has a faster learning speed than the conventional CMAC, but also provides output derivatives and more precise learning result. Besides, compared with other enhanced CMAC models providing output derivatives, the proposed method has fastest learning speed.",
author = "Chen, {Chih Ming} and Hong, {Chin Ming}",
year = "2001",
language = "English",
volume = "2",
pages = "1009--1014",
editor = "M.H. Smith and W.A. Gruver and L.O. Hall",
booktitle = "Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS",

}

TY - GEN

T1 - A weighted grey CMAC neural network with output differentiability

AU - Chen, Chih Ming

AU - Hong, Chin Ming

PY - 2001

Y1 - 2001

N2 - The Cerebellar Model Arithmetic Computer (CMAC) is a table lookup neurocomputing technique. It can be viewed as a basis function network (BFN) and performs well in terms of its fast learning speed, local generalization capability for approximating nonlinear functions. However, a disvantage is that the derivative of its output cannot be preserved due to the CMAC uses a constant basis function within each quantized state. This creates the limitation and inconvenience while the derivative information is needed in real-world applications. This paper proposes a Weight Grey CMAC (WGCMAC) that includes the conventional CMAC weight addressing scheme and the weighted grey prediction model to resolve this problem. Based on the weighted grey prediction model, we present an efficient learning algorithm for the proposed WGCMAC. Experiments confirm that the WGCMAC not only has a faster learning speed than the conventional CMAC, but also provides output derivatives and more precise learning result. Besides, compared with other enhanced CMAC models providing output derivatives, the proposed method has fastest learning speed.

AB - The Cerebellar Model Arithmetic Computer (CMAC) is a table lookup neurocomputing technique. It can be viewed as a basis function network (BFN) and performs well in terms of its fast learning speed, local generalization capability for approximating nonlinear functions. However, a disvantage is that the derivative of its output cannot be preserved due to the CMAC uses a constant basis function within each quantized state. This creates the limitation and inconvenience while the derivative information is needed in real-world applications. This paper proposes a Weight Grey CMAC (WGCMAC) that includes the conventional CMAC weight addressing scheme and the weighted grey prediction model to resolve this problem. Based on the weighted grey prediction model, we present an efficient learning algorithm for the proposed WGCMAC. Experiments confirm that the WGCMAC not only has a faster learning speed than the conventional CMAC, but also provides output derivatives and more precise learning result. Besides, compared with other enhanced CMAC models providing output derivatives, the proposed method has fastest learning speed.

UR - http://www.scopus.com/inward/record.url?scp=0035792606&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0035792606&partnerID=8YFLogxK

M3 - Conference contribution

VL - 2

SP - 1009

EP - 1014

BT - Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS

A2 - Smith, M.H.

A2 - Gruver, W.A.

A2 - Hall, L.O.

ER -