Spectral analysis and synthesis of three-layered feed-forward neural networks for function approximation

Author

Pelagotti, Andrea ; Piuri, Vincenzo

Author_Institution

Dept. of Electron. & Inf., Politecnico di Milano, Italy

fYear

1996

fDate

12-14 Feb 1996

Firstpage

239

Lastpage

245

Abstract

The universal approximation capability exhibited by one-hidden-layer neural network is explored to create a new synthesis method for minimized architectures suited for VLSI implementation. The development is based on the spectral analysis of the network, which focuses their capability of combining single neurons spectra to obtain the spectrum of the function to approximate. In this paper, we propose a new spectrum-based technique to synthesize 1-N-1 networks which approximate y=f(x) functions, with x, y∈R

Keywords

VLSI; feedforward neural nets; function approximation; learning (artificial intelligence); multilayer perceptrons; neural chips; spectral analysis; 1-N-1 networks; VLSI implementation; function approximation; minimized architectures; one-hidden-layer neural network; single neurons spectra; spectral analysis; synthesis method; three-layered feed-forward neural networks; Feedforward neural networks; Feedforward systems; Frequency; Function approximation; Minimization; Network synthesis; Neural networks; Neurons; Spectral analysis; Very large scale integration;

fLanguage

English

Publisher

ieee

Conference_Titel

Microelectronics for Neural Networks, 1996., Proceedings of Fifth International Conference on

Conference_Location

Lausanne

ISSN

1086-1947

Print_ISBN

0-8186-7373-7

Type

conf

DOI

10.1109/MNNFS.1996.493797

Filename

493797