Access control in MI Machine Learning

The MI Machine Learning apps do not currently recognize any role-based access control you may have implemented in your Granta MI system (such as restricting a user's access to specific databases, tables or data).

All users are able to select from all databases and tables when creating a new model. While access control will determine which records users can add to a Favorites list, and therefore add to a new model, once those records have been added to a model they are visible to all users in MI Machine Learning:

• The names of these records can be viewed by any user from the Populate Matrix tab in the Training app, but the datasheet can only be viewed by users who would be able to see it elsewhere.
• All users can also clone any model to use the tables, records and attributes assigned to it in their own model, even if they would not usually have access to that data, and select the model and its results in Visualize and Optimize.