Machine Learning

This feature is currently in Early Access Preview and only available for enabled customers. If you are interested in trying out this feature, please visit neo4j.com.

Aura Graph Analytics can be used to create, store and reuse machine learning models created by one of the available machine learning algorithms.

The Model Catalog is a persistent store for trained machine learning models. Once stored, a model can be reused in later sessions, even after the original GDS Session has ended.

Models are scoped by:

  • User: A model is owned by the user that created the session responsible for storing it.

  • Project: GDS Sessions can access models managed within the same Aura project.

  • Cloud provider + region: Models are only available to Sessions in the same cloud provider and region where they were stored.

Usage

The model catalog can be managed by the following model catalog operations.

Access to the model catalog via the Cypher API for Aura Graph Analytics requires an additional parameter sessionName.

Limitations

Model storage in Aura Graph Analytics has some limitations.

  • The gds.model.publish procedure is currently not supported, so models cannot be shared with other users.

  • Models cannot be accessed across regions or cloud providers.