Kernekoncepter
The Bipartite Mixed Membership Distribution-Free (BiMMDF) model allows for community detection in overlapping bipartite weighted networks where the edge weights can follow any distribution.
Resumé
The content introduces a novel model called the Bipartite Mixed Membership Distribution-Free (BiMMDF) model for community detection in overlapping bipartite weighted networks.
Key highlights:
BiMMDF allows the adjacency matrix elements to follow any distribution as long as the expectation matrix has a block structure related to node membership.
BiMMDF can model overlapping bipartite signed networks and is an extension of previous models like the mixed membership stochastic blockmodels.
An efficient spectral algorithm with theoretical guarantees is proposed to fit the BiMMDF model.
The separation conditions for BiMMDF are derived for different distributions, showing the differences in requirements.
The model is also extended to handle missing edges in sparse networks.
Extensive experiments on synthetic and real-world networks demonstrate the advantages of the BiMMDF model.
Statistik
The content does not contain any specific metrics or figures to support the key logics. It is a methodological paper introducing a novel model.
Citater
The content does not contain any striking quotes to support the key logics.