Graph Embeddings: How nodes get mapped to vectors.

Graph Embeddings: How nodes get mapped to vectors

• Most traditional Machine Learning Algorithms work on numeric vector data
• Graph embeddings unlock the powerful toolbox by learning a mapping from graph structured data to vector representations
• Their fundamental optimization is: Map nodes with similar contexts close in the embedding space
• The context of a node in a graph can be defined using one of two orthogonal approaches — Homophily and Structural Equivalence — or a combination of them
• Once the metrics …

data science graph graph-data-science graph-embedding machine learning