Open mblohr opened 9 years ago
Since graphs of a given number of labeled nodes from a discrete (indeed, finite) space, it seems to me you could definitely compute KL divergence between two graph-valued random variables, where the sum is over each possible graph in the space, and the probabilities are the pmfs of the individual graph-valued RVs in question.
Is Statistical Connectomics related in some way to random graph theory? Would it be possible to calculate a Kullback-Leibler or other type of divergence between two graphs to compare them?