Chapter 5. Deploying Automated Machine Learning Models

Microsoft Azure Machine Learning enables you to manage the life cycle of your machine learning models. After you have trained the models by using Azure Machine Learning’s automated ML tool, you can retrieve the best model identified, and register the model with Azure Machine Learning. Model registration enables you to store varying versions of models in the machine learning workspace and makes it possible for you to easily deploy the models to different target environments.

In this chapter, we explore how to use Azure Machine Learning to do the following:

  • Register the best model produced by automated ML.

  • Specify and develop the scoring file. The scoring will be included as part of the container images that will be generated.

  • Deploy the models to Microsoft Azure Container Instances (ACI) and Azure Kubernetes Service (AKS).

  • Troubleshoot failures during model and web service deployments.

Deploying Models

In Chapter 3, you learned how to build a machine learning model using automated ML. In this section, you’ll learn how to register and deploy the best model that is identified by automated ML. Azure Machine Learning supports a rich set of deployment environments, ranging from REST APIs hosted in Azure, to models deployed to different edge devices and hardware. These environments include the following:

  • Azure Machine Learning Compute

  • ACI

  • AKS

  • Azure IoT Edge


To learn more about the up-to-date list of deployment ...

Get Practical Automated Machine Learning on Azure now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.