Inference and learning in sparse systems with multiple states

A Braunstein,A Ramezanpour,R Zecchina,P Zhang
DOI: https://doi.org/10.1103/PhysRevE.83.056114
Abstract:We discuss how inference can be performed when data are sampled from the nonergodic phase of systems with multiple attractors. We take as a model system the finite connectivity Hopfield model in the memory phase and suggest a cavity method approach to reconstruct the couplings when the data are separately sampled from few attractor states. We also show how the inference results can be converted into a learning protocol for neural networks in which patterns are presented through weak external fields. The protocol is simple and fully local, and is able to store patterns with a finite overlap with the input patterns without ever reaching a spin-glass phase where all memories are lost.
What problem does this paper attempt to address?