The effects of limited-precision weights on the threshold Adaline

Author

Stevenson, Maryhelen

Author_Institution

Dept. of Electr. Eng., New Brunswick Univ., Fredericton, NB, Canada

Volume

8

Issue

3

fYear

1997

fDate

5/1/1997 12:00:00 AM

Firstpage

549

Lastpage

552

Abstract

The effect of limited-precision weights on the functional capability of a threshold Adaline is examined. The number of logic functions which can be implemented by the threshold Adaline serves as the primary measure of functional capability. Closed-form expressions are provided for the number of logic functions which can be implemented by a threshold Adaline with four different levels of weight precision. In addition, realization tests and procedures (based on the characterizing numbers) are provided for each level of weight precision. The number of realizable logic functions is compared to the capacity of a threshold Adaline with limited precision weights and a relationship between the two measures is proposed

Keywords

formal logic; learning (artificial intelligence); neural nets; functional capability; learning algorithm; limited-precision weights; neural networks; separable logic functions; threshold Adaline; threshold logic; Artificial neural networks; Closed-form solution; Computational modeling; Computer networks; Computer simulation; Digital arithmetic; Floating-point arithmetic; Logic functions; Neural network hardware; Testing;

fLanguage

English

Journal_Title

Neural Networks, IEEE Transactions on

Publisher

ieee

ISSN

1045-9227

Type

jour

DOI

10.1109/72.572094

Filename

572094