The Hopfield model on a sparse Erdös-Renyi graph (Q537898): Difference between revisions

The storage capacity of the Hopfield model on a sparse \(G(N,p)\) random graph is analyzed. The Hopfield model is well known to be a spin glass on the one hand, but also to model a neural network, on the other. It is proved that the storage capacity is proportional to \(\alpha p N\) (with \(\alpha\leq \alpha_c=0.03\)) in the entire regime where the corresponding random graph is asymptotically connected.