### Abstract

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.

Original language | English |
---|---|

Title of host publication | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |

Publisher | Springer Verlag |

Pages | 34-47 |

Number of pages | 14 |

Volume | 1011 |

ISBN (Print) | 9783540606079 |

Publication status | Published - 1995 |

Externally published | Yes |

Event | 3rd World Wisepersons Workshop on Fuzzy Logic and Neural Networks/Genetic Algorithms, WWW 1994 - Nagoya, Japan Duration: Aug 9 1994 → Aug 10 1994 |

### Publication series

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

Volume | 1011 |

ISSN (Print) | 03029743 |

ISSN (Electronic) | 16113349 |

### Other

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

Country | Japan |

City | Nagoya |

Period | 8/9/94 → 8/10/94 |

### Fingerprint

### ASJC Scopus subject areas

- Computer Science(all)
- Theoretical Computer Science

### Cite this

*Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)*(Vol. 1011, pp. 34-47). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 1011). Springer Verlag.

**Fuzzy gaussian potential neural networks using a functional reasoning.** / Teshnehlab, Mohammad; Watanabe, Keigo.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics).*vol. 1011, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 1011, Springer Verlag, pp. 34-47, 3rd World Wisepersons Workshop on Fuzzy Logic and Neural Networks/Genetic Algorithms, WWW 1994, Nagoya, Japan, 8/9/94.

}

TY - GEN

T1 - Fuzzy gaussian potential neural networks using a functional reasoning

AU - Teshnehlab, Mohammad

AU - Watanabe, Keigo

PY - 1995

Y1 - 1995

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

M3 - Conference contribution

SN - 9783540606079

VL - 1011

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

SP - 34

EP - 47

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

PB - Springer Verlag

ER -