Model Registry

The Model Registry is a collection of models that will run federated out-of-the-box on datasets that are registered to Apheris Compute Gateways. The Model Registry combines different data-driven algorithms that include statistical, machine learning or deep learning implementations.

When you start your journey on the Apheris platform, your Model Registry is pre-populated with Apheris-defined models. It is furthermore possible to build your own customized models for your collaborators and yourself. You can learn more about creating custom models in the next chapter.

When clicking an item in the model list, you are forwarded to the model view.

The list view gives you an overview of federation ready models that are available in the platform. Models are sorted by their latest version update.

Model view

The model view shows you details about:

• the author and organization that created this model;
• its first creation date;
• its versions, as well as the underlying federation framework;
• the ability to compare the source code of different versions (please contact your Apheris representative if you do not have access to the model's source code repository);
• details about its capabilities, intended use and architecture;
• recommendations regarding its security, privacy and regulatory compliance.

The model view gives you detailed information about the model metadata like author and creation date, capabilities, different versions and their underlying federation framework as well as governance, privacy, and security recommendations.