Azure Machine Learning is an integrated, end-to-end data science and advanced analytics solution. It enables data scientists to prepare data, develop experiments, and deploy models at cloud scale.
The main components of Azure Machine Learning are:
- Azure Machine Learning Workbench
- Azure Machine Learning Experimentation Service
- Azure Machine Learning Model Management Service
- Microsoft Machine Learning Libraries for Apache Spark (MMLSpark Library)
- Visual Studio Code Tools for AI
Together, these applications and services help significantly accelerate your data science project development and deployment.
Read Also: What is Machine Learning?
Azure Machine Learning Compatibility to Open Source
Azure Machine Learning fully supports open source technologies. Thousands of open source Python packages, such as the following machine learning frame works:
- scikit-learn
- TensorFlow
- Microsoft Cognitive Toolkit
- Spark ML
One can execute their experiments in managed environments such as Docker containers and Spark clusters.
Azure Machine Learning is built on top of the following open source technologies:
- Jupyter Notebook
- Apache Spark
- Docker
- Kubernetes
- Python
- Conda
It also includes Microsoft’s own open source technologies, such as Microsoft Machine Learning Library for Apache Spark and Cognitive Toolkit.
In addition, you also benefit from some of the most advanced, tried-and-true machine learning technologies developed by Microsoft designed to solve on large-scale problems. They are battle-tested in many Microsoft products, such as:
- Windows
- Bing
- Xbox
- Office
- SQL Server
The following are some of the Microsoft machine learning technologies included with Azure Machine Learning:
- PROSE (PROgram Synthesis using Examples)
- microsoftml
- revoscalepy
Azure Machine Learning Model Management
Azure Machine Learning Model Management enables you to manage and deploy machine-learning workflows and models.
Model Management provides capabilities for:
- Model Versioning
- Tracking models in production
- Deploying models to production through AzureML Compute Environment with Azure Container Service and Kubernetes
- Creating Docker containers with the models and testing them locally
- Automated model retraining
- Capturing model telemetry for actionable insights.
Azure Machine Learning Model Management provides a registry of model versions. It also provides automated workflows for packaging and deploying Machine Learning containers as REST APIs. The models and their runtime dependencies are packaged in Linux-based Docker container with prediction API.
Azure Machine Learning Compute Environments help to set up and manage scalable clusters for hosting the models. The compute environment is based on Azure Container Services. Azure Container Services provides automatic exposure of Machine Learning APIs as REST API endpoints with the following features:
- Authentication
- Load balancing
- Automatic scale-out
- Encryption
Azure Machine Learning Model Management provides these capabilities through the CLI, API, and the Azure portal.
Azure Machine Learning model management uses the following information:
- Model file or a directory with the model files
- User created Python file implementing a model scoring function
- Conda dependency file listing runtime dependencies
- Runtime environment choice, and
- Schema file for API parameters
This information is used when performing the following actions:
- Registering a model
- Creating a manifest that is used when building a container
- Building a Docker container image
- Deploying a container to Azure Container Service
The following figure shows an overview of how models are registered and deployed into the cluster.
