<ul><li>This paper introduces a regression method based on multi-output Gaussian processes (MOGP) for analyzing graph-structured data.</li><li>The method aims to capture correlations between vertices and associated data within the graph structure.</li><li>The formulation of MOGP is versatile and adaptable to various data configurations and inference scenarios, offering flexibility in kernel design.</li><li>Experimental evaluation using synthetic and real data demonstrates the performance and potential extensions of the proposed MOGP formulation.</li></ul>

Multi-Output Gaussian Processes for Graph-Structured Data

Discover more