logo

Graph Embedding, Node Embedding, Edge Embedding 📂Graph Theory

Graph Embedding, Node Embedding, Edge Embedding

Definition1

Let the graph G(V,E)G(V, E) be given. The function f:VRnf: V \to \mathbb{R}^{n} is called node embedding, and the function g:ERmg: E \to \mathbb{R}^{m} is called edge embedding.

For the set of graphs G={Gi}\mathcal{G} = \left\{ G_{i} \right\}, h:GRkh: \mathcal{G} \to \mathbb{R}^{k} is referred to as graph embedding.

Explanation

Graph/node/edge embedding is a function that maps a graph or its constituents to Euclidean space. To handle the abstract entity of a graph as data,

these functions are often referred to by different names. Node embedding is termed as graph signal in the field of graph signal processing, and the terms graph/node/edge feature are also used.