AI-900 Question 138
Select 3You are a data scientist working on a machine learning project using Azure Machine Learning. You have successfully trained and registered your model in the Azure Machine Learning workspace. What are the next steps to deploy this model as a web service and manage it effectively?
- A
Create an inference configuration specifying the scoring script and environment dependencies.
- B
Manually configure a virtual machine for hosting the model and deploy it using Azure CLI.
- C
Deploy the model to an endpoint, such as an Azure Kubernetes Service (AKS) cluster or Azure Container Instances (ACI).
- D
Use Azure Machine Learning to monitor the deployed model for performance and data drift.
- E
Export the model to a local machine for testing and redeploy it manually if needed.
Show answer and explanation
Correct answers: A, C, D
Explanation
Azure Machine Learning provides comprehensive tools and features for deploying and managing machine learning models. To deploy a model, you need to create an inference configuration, deploy it to a suitable compute target such as AKS or ACI, and use Azure Machine Learning’s monitoring tools to ensure the model performs well in production. Manual steps like virtual machine configuration or local testing are not aligned with Azure's streamlined deployment and management capabilities.
- A. Correct.
Correct. Creating an inference configuration is essential to specify how the model will process incoming requests and the required runtime environment for deployment.
- B. Incorrect.
Incorrect. While you can manually configure infrastructure, Azure Machine Learning provides streamlined deployment options, such as AKS or ACI, eliminating the need for manual configurations.
- C. Correct.
Correct. Deploying the model to an endpoint like AKS or ACI is one of the core deployment capabilities provided by Azure Machine Learning.
- D. Correct.
Correct. Azure Machine Learning offers tools to monitor deployed models for performance issues and data drift, which are critical for effective model management.
- E. Incorrect.
Incorrect. Exporting the model to a local machine for testing is not a standard or recommended step in Azure Machine Learning deployment workflows.