Premium
A Projection Rule for Complex‐Valued Associative Memory with Partial Connections
Author(s) -
Tsuji Masayuki,
Isokawa Teijiro,
Kobayashi Masaki,
Matsui Nobuyuki,
Kamiura Naotake
Publication year - 2020
Publication title -
ieej transactions on electrical and electronic engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.254
H-Index - 30
eISSN - 1931-4981
pISSN - 1931-4973
DOI - 10.1002/tee.23200
Subject(s) - embedding , content addressable memory , associative property , computer science , projection (relational algebra) , hopfield network , learning rule , scheme (mathematics) , bidirectional associative memory , artificial intelligence , rule based system , algorithm , theoretical computer science , artificial neural network , mathematics , pure mathematics , mathematical analysis
Projection rule is a powerful learning scheme for embedding patterns onto Hopfield‐type associative memories, but this cannot be used for non‐fully connected networks, such as bidirectional associative memory. This paper presents a novel type of Projection rule that can embed patterns onto partially connected networks. The applicability and performances of the proposed Projection rule are evaluated through numerical experiments. From experimental results, it is confirmed that the proposed scheme actually works on partially connected networks, though the embedding capability is not high with compared to the one in fully connected networks. It is also shown that gray‐scaled images can be embedded and retrieved in the networks. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.