Jump to content

Talk:Stochastic block model

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

A little to-do list

  • It'd be great to have plots for the impossibility thresholds, maybe marked with text "possible" / "impossible"
  • Overall this article is probably a little dense. Plots as above would help a little. But community structure is always a slightly lighter introduction; maybe we can point there for that purpose.
  • It'd be nice to expand the Algorithms section. Cite Hajek–Wu–Xu, Agarwal et al for SDP approaches; mention this "no-return random walks" idea from Massoulié to overcome spectral noise in constant-average-degree graphs, and also maybe cover some of the less recent techniques (e.g. Girvan–Newman modularity), though those are covered somewhat in community structure.
  • Say more about variants: censored block model, mixed-membership models, etc. These could each get a section.
  • Worth describing the statistical thresholds as being like the giant component / connectivity thresholds in ER random graphs, but with some oversampling?

Will Perry (talk) 14:55, 22 September 2015 (UTC)[reply]

Added image

I have added a self-created image of an assortative case of the stochastic block model with two communities. The image has also been uploaded to Wikimedia Commons. Rosey9921 (talk) 04:04, 20 April 2021 (UTC)[reply]