TY - GEN

T1 - Fuzzy gaussian potential neural networks using a functional reasoning

AU - Teshnehlab, Mohammad

AU - Watanabe, Keigo

PY - 1995/1/1

Y1 - 1995/1/1

N2 - This paper presents the principal design of a fuzzy gaussian potential neural network (FGPNN) to achieve high capability to learn expert control rules of the fuzzy controller. In this construction, each membership function consists of a gaussian potential function (GPF) which causes the utilization of a reduced number of labels, and eventually the complexity of structural design becomes simple, specially for large scale inputs, This in turn reduces the learning trials, to improve the learning speed. Thus, the time of the training process, which is based on the ba~k-propagation method, is shortened. The construction of an FGPNN is carried out with the minimum number of GPF, based on the number of input patterns, to learn the mean vectors and shapes of the individual GPFs that basically depend on the desired trajectory. Finally, we provide a simulation to evaluate the proposed method for a multi input-output, twolink manipulator.

AB - This paper presents the principal design of a fuzzy gaussian potential neural network (FGPNN) to achieve high capability to learn expert control rules of the fuzzy controller. In this construction, each membership function consists of a gaussian potential function (GPF) which causes the utilization of a reduced number of labels, and eventually the complexity of structural design becomes simple, specially for large scale inputs, This in turn reduces the learning trials, to improve the learning speed. Thus, the time of the training process, which is based on the ba~k-propagation method, is shortened. The construction of an FGPNN is carried out with the minimum number of GPF, based on the number of input patterns, to learn the mean vectors and shapes of the individual GPFs that basically depend on the desired trajectory. Finally, we provide a simulation to evaluate the proposed method for a multi input-output, twolink manipulator.

UR - http://www.scopus.com/inward/record.url?scp=84948137734&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84948137734&partnerID=8YFLogxK

U2 - 10.1007/3-540-60607-6_3

DO - 10.1007/3-540-60607-6_3

M3 - Conference contribution

AN - SCOPUS:84948137734

SN - 9783540606079

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 34

EP - 47

BT - Advances in Fuzzy Logic, Neural Networks and Genetic Algorithms - IEEE/Nagoya-University World Wisepersons Workshop, 1994, Selected Papers

A2 - Furuhashi, Takeshi

PB - Springer Verlag

T2 - 3rd World Wisepersons Workshop on Fuzzy Logic and Neural Networks/Genetic Algorithms, WWW 1994

Y2 - 9 August 1994 through 10 August 1994

ER -