Gaussian Processes on Graphs Via Spectral Kernel Learning

We propose a graph spectrum-based Gaussian process for prediction of signals defined on nodes of the graph. The model is designed to capture various graph signal structures through a highly adaptive kernel that incorporates a flexible polynomial function in the graph spectral domain. Unlike most existing approaches, we propose to learn such a spectral kernel defined on a discrete space. In addition, this kernel has the interpretability of graph filtering achieved by a bespoke maximum likelihood learning algorithm that enforces the positivity of the spectrum. We demonstrate the interpretability of the model through synthetic experiments from which we show various ground truth spectral filters can be accurately recovered, and the adaptability translates to improved predictive performances compared to the baselines on real-world graph data of various characteristics.