Skip Navigation

Azure Machine Learning service

Accelerate machine learning from the cloud to the edge

Why Azure Machine Learning service?

Productive

Increase your rate of experimentation and build models faster with automated machine learning and managed compute.

Open

Use the machine learning libraries and IDEs that you already know.

Trusted

Benefit from enterprise readiness with the security and compliance features of Azure and virtual network support.

Hybrid

Build, train, manage, and deploy machine learning models from the cloud to the edge.

What comes with Azure Machine Learning service

Automated machine learning and hyper-parameter tuning

Identify the best algorithms faster with automated machine learning, and find the best model efficiently with intelligent hyper-parameter tuning.

Version control and reproducibility

Increase your rate of experimentation by tracking and logging your experiments for reproducibility and easy modification.

Support for open source libraries and IDEs

Use machine learning libraries such as Tensorflow, PyTorch, and scikit-learn. Azure Machine Learning service integrates with your favorite Python IDE, including Visual Studio Code, Visual Studio, Azure Databricks notebooks, or Jupyter notebooks.

Model management

Proactively manage and monitor your models using the image and model registry, and upgrade them through integrated CI/CD.

Hybrid deployment

Deploy models where you need them most with managed deployments to the cloud and the edge.

Distributed deep learning

Build better models faster with massive, managed GPU clusters. Train models quickly with distributed deep learning, and deploy them on FPGAs.

How to use Azure Machine Learning service

Step 1: Create a workspace

Install the SDK and create your workspace to store your compute resources, models, deployments, and run histories in the cloud.

Step 2: Train a model

Train a model locally or in the cloud, using open source machine learning libraries. Track your experiments and easily scale up or out your training with managed compute resources in the cloud.

Step 3: Deploy and manage

Deploy your model to test or production to generate predictions. Deploy to the cloud or at the edge, or leverage hardware-accelerated models on FPGAs for super-fast inferencing. When your model is in production, monitor it for performance and data drift, and retrain it as needed.

Related products and services

Azure Databricks

Fast, easy, and collaborative Apache Spark-based analytics platform

Machine Learning Studio

Easily build, deploy, and manage predictive analytics solutions

Data Science Virtual Machines

Rich pre-configured environment for AI development

Start making better decisions using Azure Machine Learning service