On-Demand Webinar – AI Development Using Data Science VMs (DSVM), Deep Learning VMs (DLVM) & Azure Batch AI

Article
11/16/2017

This post is authored by Barnam Bora, Program Manager in the Cloud AI group at Microsoft.

Microsoft's Data Science Virtual Machines (DSVM) and Deep Learning Virtual Machines (DLVM) are a family of popular VM images in Windows Server and Linux flavors that are published on the Azure Marketplace. They have a curated but broad set of pre-configured machine learning and data science tools including pre-loaded samples. DSVM and DLVM are configured and tested to work seamlessly with a plethora of services available on the Microsoft Azure cloud, and they enable a wide array of data analytics scenarios that are being used by many organizations across the globe.

We recently hosted a webinar covering the workflow of building ML and AI -powered solutions in Azure using DSVM, DLVM and related services such as Azure Batch AI and Azure Machine Learning Model Management. The webinar video is available from the link below (requires registration with Microsoft) and more information about the webinar are in the sections that follow.

WATCH: AI development using DSVM/DLVM

Scenarios Covered in the Webinar

Single GPU Node AI Model Training

DSVM and DLVM are great tools to develop, test and deploy AI models and solutions. Data scientists and developers can use the capabilities provided in DSVM/DLVM to start developing AI solutions on a single node/machine. Once initial development is complete and there's a need to train on a much larger dataset, it's remarkably simple to scale out from a single node to a multi-node scalable cluster for parallelized training of models using the Azure Batch AI service.

Scale Out AI Model Training with DSVM and DLVM on Azure Batch AI

This section is a detailed discussion and demonstration of using the DSVM/DLVM for single node development and testing of AI Models and then scaling out to a multi-node cluster using the Azure Batch AI Service. The dataset used for this sample is the CIFAR-10 Dataset.

AI Model Deployment & Management using Azure Machine Learning Model Management

In this section we discuss the workflow for taking a trained model and packaging and deploying it using Azure Machine Learning Model Management to build operational scoring pipelines.

Other Sections Covered in the Webinar

Introduction to the Data Science & Deep Learning Virtual Machines (DSVM/DLVM) offerings in Azure

This section introduces the DSVM and DLVM offerings and familiarizes participants with the functionality that they include.

The Typical AI Development Workflow
This section is a discussion about the typical end-to-end AI solution development workflow and how DSVM/DLVM enable data scientists and developers to build AI solutions.