Train a machine learning model with Azure Machine Learning

Question 1

What is an Estimator in Microsoft Azure?

Answer 1

In Azure Machine Learning, you can use a Run Configuration and a Script Run Configuration to run a script-based experiment that trains a machine learning model. However, depending on the machine learning framework being used and the dependencies it requires, the run configuration may become complex. Azure Machine Learning also provides a higher level abstraction called an Estimator that encapsulates a run configuration and a script configuration in a single object, and for which there are pre-defined, framework-specific variants that already include the package dependencies for common machine learning frameworks such as Scikit-Learn, PyTorch, and Tensorflow

Question 2

What are parameters for in the context of estimators

Answer 2

You can increase the flexibility of script-based experiments by using parameters to set variables in the script. Meaning that when specifying a model, it is possible to afterwards change several parameters of the model when deploying

Question 3

How is parameter setting done in Microsoft Azure? Does it need a library etc?

Answer 3

It requires the argparse library. After the model is saved, it is possible to assign a value to a parameter when the model is deployed

Question 4

What is model registration for?

Answer 4

Model registration enables you to track multiple versions of a model, and retrieve models for inferencing (predicting label values from new data). When you register a model, you can specify a name, description, tags, framework, framework version, custom properties, and other useful metadata

Question 5

You have run an experiment to train a model. You want the model to be stored in the workspace, and available to other experiments and published services. What should you do?

Answer 5

Register the model in the workspace