Graph Powered Machine Learning From Graph Analytics to Graph AI | Jörg Schad | Connected Data World

This master class podcast dives into the exciting intersection of graph-powered machine learning, showcasing how to integrate graph analytics with machine learning techniques. It begins with a quick overview of essential graph concepts and databases, then moves on to explore graph analytics, including key algorithms like PageRank and centrality measures. The discussion culminates in graph machine learning, where the focus is on graph embeddings and Graph Neural Networks. A standout example shared in the podcast is Google’s use of graph machine learning to enhance estimated time of arrival predictions on Google Maps. The main takeaway is that grasping the relationships between data points, as illustrated through graphs, can greatly enhance the accuracy of machine learning models and unlock valuable insights across various fields, from recommendation engines to supply chain management. The presenter also highlights the importance of selecting the appropriate approach—whether graph analytics or graph machine learning—tailored to the specific challenges and attributes of the dataset at hand.