Training

Construct a feature matrix and define training and test datasets from data stored in Granta MI, then send the untrained model to the Intellegens Alchemite engine for training and testing.

In MI Machine Learning, the process for defining and training a model is as follows:
  1. Create a new untrained model, or clone an existing trained or untrained model.
  2. Choose a Granta MI database and primary table.
  3. Define and populate the model Matrix. A Matrix contains attributes, links and records from Granta MI, and is converted into a feature matrix before being sent to the engine.
  4. Send the model to the engine for training, and track progress with the Job Queue.
After your model is trained, you can visualize the model, or apply a set of constraints (for example, optimizing an additive manufacturing process to produce materials with the desired properties).