The paper states thats the Exchangeable Graph Model can represent any statistical model for a graph (p25) which makes sense give a parametric model (one with a fixed length for K) but what about non-parametric models (where K could be infinite)?
Does this imply that there isn't such things as non-parametric models for graphs?
(Edit: Scratch this last part, just did a google search and there are papers on non-parametric models for graphs)
The paper states thats the Exchangeable Graph Model can represent any statistical model for a graph (p25) which makes sense give a parametric model (one with a fixed length for K) but what about non-parametric models (where K could be infinite)?
Does this imply that there isn't such things as non-parametric models for graphs? (Edit: Scratch this last part, just did a google search and there are papers on non-parametric models for graphs)