Associative memory neural networks for error correction of linear block codes

مرکز و کتابخانه مطالعات اسلامی به زبان های اروپایی

منو

" Associative memory neural networks for error correction of linear block codes "
M. A. Sayani

Document Type	:	Latin Dissertation
Language of Document	:	English
Record Number	:	1113388
Doc. No	:	TLpq231443539
Main Entry	:	M. A. Sayani
Title & Author	:	Associative memory neural networks for error correction of linear block codes\ M. A. Sayani
College	:	King Fahd University of Petroleum and Minerals (Saudi Arabia)
Date	:	1995
student score	:	1995
Degree	:	M.S.
Page No	:	105
Abstract	:	Associative memory neural networks are used for error correction of linear block codes. The implementation of decoder based on neural networks does not require any special characteristics of codes (i.e., Linearity, cyclic nature etc.) and can decode many different types of codes such as repetition, Hamming, BCH, RS, and other codes. The concept of Hopfield model has been applied for error correction of linear block codes defined over GF(q) fields. All the codewords of length n are considered as stable states which are used to construct the weight matrix as defined in the Hopfield model. All the other possible words of length n are the unstable states. For a linear (n, k) code, the number of stable states are 2 and the possible number of unstable states (patterns) are 2. The decoder would either map the unstable state to one of the stable states or indicates that an error has occurred. The error correction capability is the same as that of classical decoding methods, that is, only limited by the minimum distance constraints. Error correction is applied for the codes having single and multiple error correction capability.
Subject	:	Applied sciences
	:	Artificial intelligence
	:	Electrical engineering