Time-varying Graph Learning Under Structured Temporal Priors
This work addresses graph learning for dynamic systems where temporal dependencies extend beyond immediate neighbors, which is relevant for applications like network analysis and time-series modeling.
The paper tackles the problem of learning time-varying graphs by introducing structured temporal priors that capture underlying relations between arbitrary graphs in a sequence, rather than just consecutive ones. The proposed method demonstrates superior performance in numerical experiments compared to existing chain structure-based approaches.
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 chain structure based methods in which the priors like temporal homogeneity can only describe the variations of two consecutive graphs, we propose a structure named \emph{temporal graph} to characterize the underlying real temporal relations. Under this framework, the chain structure is actually a special case of our temporal graph. We further proposed Alternating Direction Method of Multipliers (ADMM), a distributed algorithm, to solve the induced optimization problem. Numerical experiments demonstrate the superiorities of our method.