1 paper · 1 filter
Jan Niklas Böhm, Marius Keute, Alica Guzmán +3
Node embeddings are a paradigm in non-parametric graph representation learning, where graph nodes are embedded into a given vector space to enable downstream processing. State-of-t…