Accelerated Graph Learning from Smooth Signals
This work addresses the challenge of efficiently learning graph structures from smooth data, which is incremental as it improves upon existing solvers with better convergence guarantees and speed.
The paper tackles the problem of network topology identification from smooth nodal signals by developing a fast dual-based proximal gradient algorithm that efficiently solves a strongly convex,regularized inverse problem, demonstrating accurate recovery of graphs significantly faster than state-of-the-art methods without extra computational cost.
We consider network topology identification subject to a signal smoothness prior on the nodal observations. A fast dual-based proximal gradient algorithm is developed to efficiently tackle a strongly convex, smoothness-regularized network inverse problem known to yield high-quality graph solutions. Unlike existing solvers, the novel iterations come with global convergence rate guarantees and do not require additional step-size tuning. Reproducible simulated tests demonstrate the effectiveness of the proposed method in accurately recovering random and real-world graphs, markedly faster than state-of-the-art alternatives and without incurring an extra computational burden.