Data fusion reconstruction of spatially embedded complex networks

Abstract We introduce a kernel Lasso (kLasso) approach which is type of sparse optimization that simultaneously accounts for spatial regularity and structural sparsity to reconstruct spatially embedded complex networks from time-series data about nodal states. Through the design function motivated by real-world network features, proposed kLasso exploits embedding distances penalize overabundance long-distance connections. Examples both random geometric graphs transportation show method improves significantly upon existing reconstruction techniques mainly concern but not regularity. Our results highlight promise information fusion in networks, utilizing microscopic node-level dynamics (e.g. time series data) macroscopic network-level (metadata or other prior information).