Time-Varying Graph Learning Under Structured Temporal Priors

This paper endeavors to learn time-varying graphs by using structured temporal priors that assume underlying relations between arbitrary two graphs in the graph sequence. Different from many existing methods that only describe vari-ations between two consecutive graphs, we propose a structure named temporal graph to characterize the underlying real temporal relations. Under this framework, classic priors like temporal homogeneity are actually special cases of our temporal graph. To address computational issue, we further develop a distributed algorithm based on Alternating Direction Method of Multipliers (ADMM) to solve the induced optimization problem. Numerical experiments on synthetic and real data demonstrate the superiorities of our method.