We propose a method for graph learning from spatiotemporal measurements. We aim at inferring time-varying graphs under the assumption that changes in graph topology and weights are sparse in time. The ...