Detail publikačního výsledku

Design of Artificial Neural Network for Antenna Synthesis using the Optimization with Variable Number of Dimensions

KADLEC, P.

Originální název

Design of Artificial Neural Network for Antenna Synthesis using the Optimization with Variable Number of Dimensions

Anglický název

Design of Artificial Neural Network for Antenna Synthesis using the Optimization with Variable Number of Dimensions

Druh

Stať ve sborníku v databázi WoS či Scopus

Originální abstrakt

Neural networks are used extensively these days to solve many problems of daily life. The architecture of the neural network has a major influence on the quality of the estimate obtained with the help of the neural network. Nevertheless, the design of the neural network architecture is far from being a solved problem. This paper formulates the Feedforward Network architecture design as a multi–objective optimization problem with a variable number of dimensions. The purpose of the neural network is to estimate the design parameters of the circularly–polarized patch antenna. An example instance of the problem is then solved with the help of the VNDGDE3 algorithm that can solve the general multi–objective optimization problem with a variable number of dimensions. Results of this problem show that a neural network with an optimal complexity can be designed with the proposed methodology.

Anglický abstrakt

Neural networks are used extensively these days to solve many problems of daily life. The architecture of the neural network has a major influence on the quality of the estimate obtained with the help of the neural network. Nevertheless, the design of the neural network architecture is far from being a solved problem. This paper formulates the Feedforward Network architecture design as a multi–objective optimization problem with a variable number of dimensions. The purpose of the neural network is to estimate the design parameters of the circularly–polarized patch antenna. An example instance of the problem is then solved with the help of the VNDGDE3 algorithm that can solve the general multi–objective optimization problem with a variable number of dimensions. Results of this problem show that a neural network with an optimal complexity can be designed with the proposed methodology.

Klíčová slova

Artificial Neural Network, Differential Evolution, Multi-objective Optimization, Variable Number of Dimensions, Antenna Design, Circular Polarization.

Klíčová slova v angličtině

Artificial Neural Network, Differential Evolution, Multi-objective Optimization, Variable Number of Dimensions, Antenna Design, Circular Polarization.

Autoři

KADLEC, P.

Rok RIV

2023

Vydáno

22.04.2022

Nakladatel

Institute of Electrical and Electronics Engineers Inc.

ISBN

978-1-7281-8686-3

Kniha

Proceedings of 32nd International Conference Radioelektronika, RADIOELEKTRONIKA 2022

Strany od

149

Strany do

154

Strany počet

6

URL

https://ieeexplore.ieee.org/document/9764945

BibTex

@inproceedings{BUT178293,
  author="Petr {Kadlec}",
  title="Design of Artificial Neural Network for Antenna Synthesis using the Optimization with Variable Number of Dimensions",
  booktitle="Proceedings of 32nd International Conference Radioelektronika, RADIOELEKTRONIKA 2022",
  year="2022",
  pages="149--154",
  publisher="Institute of Electrical and Electronics Engineers Inc.",
  doi="10.1109/RADIOELEKTRONIKA54537.2022.9764945",
  isbn="978-1-7281-8686-3",
  url="https://ieeexplore.ieee.org/document/9764945"
}