Computing and Information Technology Interactive Digital Educational Library

 

CITIDEL >
Computer Science Teaching Center >
Computer Science Teaching Center >

Please use this identifier to cite or link to this item: http://hdl.handle.net/10117/186

Title: Neural Network Associative Memories and Input Space With Non-Orthonormal Basis
Authors: Suarez, Joel
Keywords: Computing Methodologies sub_subject: General Computing Methodologies sub_subject: Image Processing and Computer Vision Computing Methodologies sub_subject: Pattern Recognition Association, Basis, Neural network, Perfect matching
Issue Date: 26-Jul-2002
Publisher: Universidad Autnoma de Hidalgo; Centro de Investigacin en Tecnologas de Informacin y Sistemas
Abstract: Based in theoretical considerations only, this paper addresses the problem of getting rid cross-talk terms en the final recalling of an autoassociative neural network when input patterns belong to a subspace embedded into the whole input pattern space. One of the methods obtained also considers optimization in matrix weight size, at the expense of redefining the weight matrix, the activation function and the output space
URI: http://www.citidel.org/handle/10117/186
Other Identifiers: 280
Appears in Collections:Computer Science Teaching Center

Files in This Item:

File Description SizeFormat
ACMCoverLetter.pdf18KbAdobe PDFView/Open
abstract.txt0KbTextView/Open
ACMCoverLetter.pdf.description0KbUnknownView/Open
dublin_core.xml1KbXMLView/Open
ACMTemp.pdf205KbAdobe PDFView/Open
metadata.txt1KbTextView/Open
ACMTemp.pdf.description0KbUnknownView/Open

All items in DSpace are protected by copyright, with all rights reserved.

 

Valid XHTML 1.0! DSpace Software Copyright © 2002-2006 MIT and Hewlett-Packard - Feedback