Fig 6. Diffusion of probability from a starting node and a diffusion-based encoding algorithm.
(A) Node color denotes a gradient of probability inherited from the diffusion of probability from the current node (in red). The node with the yellow star denotes the node which inherited the highest probability. (B) A set is assigned higher probability (PA) if it is highly connected and assigned lower probability if it is sparsely connected in the graph. The arrows point to the node the diffusion-based node ranking algorithm ranked highest, based on diffusion of probability from the previous encoded node. While the p-value of the most connected node set of size 2 for the graph illustrated here is 0.25, with larger graphs and larger subset sizes, the potential for further compression increases, and thus, more significant p-values are possible.