Thouless-Anderson-Palmer equations for neural networks

Citation
M. Shamir et H. Sompolinsky, Thouless-Anderson-Palmer equations for neural networks, PHYS REV E, 61(2), 2000, pp. 1839-1844
Citations number
9
Categorie Soggetti
Physics
Journal title
PHYSICAL REVIEW E
ISSN journal
1063651X → ACNP
Volume
61
Issue
2
Year of publication
2000
Pages
1839 - 1844
Database
ISI
SICI code
1063-651X(200002)61:2<1839:TEFNN>2.0.ZU;2-#
Abstract
Previous derivation of the Thouless-Anderson-Palmer (TAP) equations for the Hopfield model by the cavity method yielded results that were inconsistent with those of the perturbation theory as well as the results derived by th e replica theory of the model. Here we present a derivation of the TAP equa tion for the Hopfield model by the cavity method and show that it agrees wi th the form derived by perturbation theory. We also use the cavity method t o derive TAP equations for the pseudoinverse neural network model. These eq uations are consistent with the results of the replica theory of these mode ls.