This is the authors' abstract. We don't add key points for this paper.
Comparing structured objects such as graphs is a fundamental operation in many learning tasks. To this end, the Gromov-Wasserstein (GW), based on Optimal Transport (OT), has proven to be successful in the specific nature of the associated objects. More specifically, the nodes connectivity relations, GW operates on graphs, seen as measures over specific spaces. At the core of OT is the idea of of mass, which imposes a coupling between all the nodes from the considered graphs. We argue in this paper that this property can be for tasks such as graph dictionary or partition learning, and we it by proposing a new semi-relaxed Gromov-Wasserstein divergence. Aside immediate computational benefits, we discuss its properties, and show that can lead to an efficient graph dictionary learning algorithm. We empirically its relevance for complex tasks on graphs such as partitioning, and completion.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Vincent-Cuaz et al. (2021) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: