Machine Learning

Question

My company provides analytics and machine learning services to enterprises. Now that we are moving over to Azure (from AWS and on-prem), I would like to understand how our workflow might change. The main aspects I am uncertain about is how we ingest data from our customers and how we deploy models for real-time and batch inference with minimal technical overheads.

Answer 1

Hi,

Thanks for reaching out to us. For real-time machine learning deployment, we now have Azure Machine Learning Designer which is very easy to use and friendly for new user.

First I want to share the workflow of Azure Machine Learning Service:

Workflow
The machine learning model workflow generally follows this sequence:

Train

Develop machine learning training scripts in Python, R, or with the visual designer.
Create and configure a compute target.
Submit the scripts to a configured compute target to run in that environment. During training, the scripts can read from or write to datastores. The logs and output produced during training are saved as runs in the workspace and grouped under experiments.
Package - After a satisfactory run is found, register the persisted model in the model registry.

Validate - Query the experiment for logged metrics from the current and past runs. If the metrics don't indicate a desired outcome, loop back to step 1 and iterate on your scripts.

Deploy - Develop a scoring script that uses the model and Deploy the model as a web service in Azure, or to an IoT Edge device.

Monitor - Monitor for data drift between the training dataset and inference data of a deployed model. When necessary, loop back to step 1 to retrain the model with new training data.

And Azure Machine Learning pipelines can help you make everything automatically: https://learn.microsoft.com/en-us/azure/machine-learning/concept-azure-machine-learning-architecture#ml-pipelines

I have shared a basic workflow and quick start here for your reference: https://learn.microsoft.com/en-us/azure/machine-learning/tutorial-designer-automobile-price-deploy

Regards,
Yutong