Neural networks with hierarchically correlated patterns

H. Gutfreund*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

71 Scopus citations

Abstract

The Hopfield model of neural networks is extended to allow for the storage and retrieval of hierarchically correlated patterns. The overlaps between these patterns form an ultrametric tree. Intermediate states, which serve as ancestors for the following levels, are generated at each level of the tree. The states belonging to each level are stored, by a modified learning rule, in a series of identical networks, one for each level. The retrieval of a particular pattern is preceded and assisted by the successive retrieval of its ancestors. The performance of this scheme is studied analytically and numerically.

Original languageEnglish
Pages (from-to)570-577
Number of pages8
JournalPhysical Review A
Volume37
Issue number2
DOIs
StatePublished - 1988

Fingerprint

Dive into the research topics of 'Neural networks with hierarchically correlated patterns'. Together they form a unique fingerprint.

Cite this