METHOD FOR GENERATING PARTIALLY CORRELATED VECTOR SETS FOR NEURAL-NETWORK SIMULATIONS

Authors
KWAN HK WANG ZW SOLTIS JJ
Citation
Hk. Kwan et al., METHOD FOR GENERATING PARTIALLY CORRELATED VECTOR SETS FOR NEURAL-NETWORK SIMULATIONS, International journal of electronics, 74(4), 1993, pp. 523-528
Citations number
6
Categorie Soggetti
Engineering, Eletrical & Electronic
Journal title
International journal of electronics → ACNP
ISSN journal
00207217
Volume
74
Issue
4
Year of publication
1993
Pages
523 - 528
Database
ISI
SICI code
0020-7217(1993)74:4<523:MFGPCV>2.0.ZU;2-B
Abstract
A simple method for generating a partially correlated and normalized c omplete vector set containing real vectors of dimensions 1 x N is prop osed. The vector sets thus obtained can be used as benchmark problems for testing the ability of a neural network to recognize partially cor related vectors (of specified cross-correlation between any two vector s) in real-world application problems.