A design method for multilayer feedforward neural networks for simple hardware implementation

Author

Kwan, Hon Keung ; Tang, Chuan Zhang

Author_Institution

Dept. of Electr. Eng., Windsor Univ., Ont., Canada

fYear

1993

fDate

3-6 May 1993

Firstpage

2363

Abstract

A method for designing a multiplierless multilayer feedforward neural network for continuous input-output mapping is presented. This method uses the simplified sigmoid activation functions at the weights in the output layer, 3-level discrete quantization functions at the hidden neurons, and single powers-of-two weights in the input layer. When tested with noisy vectors, the multiplierless network can achieve high recall accuracy, while having increased computational speed in practical applications and reduced hardware cost in digital implementation

Keywords

feedforward neural nets; multilayer perceptrons; quantisation (signal); computational speed; continuous input-output mapping; digital implementation; hardware implementation; hidden neurons; multilayer feedforward neural networks; noisy vectors; output layer; recall accuracy; simplified sigmoid activation functions; single powers-of-two weights; three-level discrete quantization; Computer networks; Design methodology; Feedforward neural networks; Hardware; Multi-layer neural network; Neural networks; Neurons; Noise reduction; Quantization; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 1993., ISCAS '93, 1993 IEEE International Symposium on

Conference_Location

Chicago, IL

Print_ISBN

0-7803-1281-3

Type

conf

DOI

10.1109/ISCAS.1993.394238

Filename

394238