Using a Feature Storage for Graph Neural Network Training
AI/Machine Learning • March 2022
Using a Feature Storage for Graph Neural Network Training
About
Using a Feature Storage for Graph Neural Network Training
About

MLOps tools are increasingly used to ensure a structured ML pipeline from which to obtain good results and improve them over time.
Feast is a Feature Store, an important component for these pipelines and we will see how to use it starting from a graph DB like Neo4j and how to load useful data to train Graph Neural Network.
The data is moved from Neo4j to Postgres to create the Offline Store for Feast, so we will talk about what the loading process can be and the associated problems up to preparing the input for GNN

Language
Italian
Level
Beginner
Length
31 minutes
Type
online conference
About the speaker
About the speaker
Valerio Piccioni
AI EngineerLarus-ba
I'm a AI Engineer focused on Graph Neural Network in Larus. I'm also an MLOps enthusiast.
Details
Language
Italian
Level
Beginner
Length
31 minutes
Type
online conference