Markov Random Geometric Graph (MRGG): A Growth Model for Temporal Dynamic Networks
This work addresses the challenge of modeling and analyzing growing temporal networks, which is incremental as it builds on latent space models but introduces a Markovian dynamic for improved temporal dependence capture.
The authors tackled the problem of modeling temporal dynamic networks by introducing Markov Random Geometric Graphs (MRGGs), a growth model based on a Markovian latent space dynamic, and provided theoretical guarantees for non-parametric estimation of its functions, with an efficient algorithm for tasks like link prediction.
We introduce Markov Random Geometric Graphs (MRGGs), a growth model for temporal dynamic networks. It is based on a Markovian latent space dynamic: consecutive latent points are sampled on the Euclidean Sphere using an unknown Markov kernel; and two nodes are connected with a probability depending on a unknown function of their latent geodesic distance. More precisely, at each stamp-time $k$ we add a latent point $X_k$ sampled by jumping from the previous one $X_{k-1}$ in a direction chosen uniformly $Y_k$ and with a length $r_k$ drawn from an unknown distribution called the latitude function. The connection probabilities between each pair of nodes are equal to the envelope function of the distance between these two latent points. We provide theoretical guarantees for the non-parametric estimation of the latitude and the envelope functions.We propose an efficient algorithm that achieves those non-parametric estimation tasks based on an ad-hoc Hierarchical Agglomerative Clustering approach. As a by product, we show how MRGGs can be used to detect dependence structure in growing graphs and to solve link prediction problems.