Azure Machine Learning
Enterprise-grade machine learning service for building and deploying models faster
Accelerate the end-to-end machine learning life cycle
Empower developers and data scientists with a wide range of productive experiences for building, training and deploying machine learning models faster. Accelerate time to market and foster team collaboration with industry-leading MLOps – DevOps for machine learning. Innovate on a secure, trusted platform, designed for responsible ML.
ML for all skills
Productivity for all skill levels – code with built-in collaborative notebooks and one-click Jupyter experience, use drag-and-drop designer or automated machine learning for accelerated model development.
End-to-end MLOps
Robust MLOps capabilities that integrate with existing DevOps processes and help manage the complete ML lifecycle.
State-of-the-art responsible ML
Responsible ML capabilities – understand models with interpretability and fairness, protect data with differential privacy and confidential computing and control the ML lifecycle with audit trials and datasheets.
Open and interoperable
Best-in-class support for open-source frameworks and languages including MLflow, Kubeflow, ONNX, PyTorch, TensorFlow, Python and R.
Boost productivity with ML for all skills
Rapidly build and deploy machine learning models using tools that meet your needs, regardless of skill level. Access built-in notebooks inside studio with a one-click Jupyter experience. Spin-up compute quickly inside notebooks and switch compute and kernels with ease. Use intellisense and code editing capabilities in notebooks and share and collaborate with your team. Use the no-code designer to get started with visual machine learning or accelerate model creation with automated machine learning, and access built-in feature engineering, algorithm selection and hyperparameter sweeping to develop highly accurate models.
Operationalise at scale with MLOps
MLOps, or DevOps for machine learning, streamlines the machine learning life cycle, from building models to deployment and management. Use ML pipelines to build repeatable workflows, and use a rich model registry to track your assets. Manage production workflows at scale by using advanced alerts and machine learning automation capabilities. Profile, validate and deploy machine learning models anywhere, from the cloud to the edge, to manage production ML workflows at scale in an enterprise-ready fashion.
Build responsible ML solutions
Access state-of-the-art responsible ML capabilities to understand protect and control your data, models and processes. Explain model behaviour during training and inferencing, and build for fairness by detecting and mitigating model bias. Preserve data privacy throughout the machine learning lifecycle with differential privacy techniques and use confidential computing to secure ML assets. Automatically maintain audit trails, track lineage and use model datasheets to enable accountability.
Innovate on an open and flexible platform
Get built-in support for open-source tools and frameworks for machine learning model training and inferencing. Use familiar frameworks like PyTorch, TensorFlow and scikit-learn, or the open and interoperable ONNX format. Choose the development tools that best meet your needs, including popular IDEs, Jupyter notebooks and CLIs – or languages such as Python and R. Use ONNX Runtime to optimise and accelerate inferencing across cloud and edge devices.
Advanced security and governance
- Get the security from the ground up and build on the trusted cloud with Azure.
- Protect access to your resources with granular role-based access, custom roles and built-in mechanisms for identity authentication.
- Build train and deploy models securely by isolating your network with virtual networks and private links.
- Manage governance with policies, audit trails, quota and cost management.
- Streamline compliance with a comprehensive portfolio spanning 60 certifications including FedRAMP High and DISA IL5.
Key service capabilities
Collaborative notebooks
Built-in notebooks with one-click Jupyter experience. Maximise productivity with intellisense, easy compute spin-up and kernel switching, and offline notebook editing.
Automated ML
Rapidly create accurate models for classification, regression and time series forecasting. Use model interpretability to understand how the model was built.
Drag and drop ML
Use designer with modules for data transformation, model training and evaluation, or to create and publish ML pipelines with a few clicks.
Data labelling
Prepare data quickly, manage and monitor labelling projects and automate iterative tasks with machine learning assisted labelling.
MLOps
Use the central registry to store and track data, models and metadata. Automatically capture lineage and governance data. Use Git to track work and GitHub Actions to implement workflows. Manage and monitor runs or compare multiple runs for training and experimentation.
Autoscaling compute
Use managed compute to distribute training and rapidly test, validate and deploy models. CPU and GPU clusters can be shared across a workspace and automatically scale to meet your ML needs.
RStudio integration
Built in R support and RStudio Server (Open-source edition) integration to build and deploy models and monitor runs.
Deep integration with other Azure services
Accelerate productivity with built-in integration with Azure services such as Azure Synapse Analytics, Cognitive Search, Power BI, Azure Data Factory, Azure Data Lake and Azure Databricks.
Reinforcement learning
Scale reinforcement learning to powerful compute clusters, support multi-agent scenarios, access open-source RL algorithms, frameworks and environments.
Responsible ML
Get model transparency at training and inferencing with interpretability capabilities. Assess model fairness through disparity metrics and mitigate unfairness. Protect data with differential privacy.
Enterprise-grade Security
Build and deploy models securely with capabilities such as network isolation and Private Link, role-based access control for resources and actions, custom roles and managed identity for compute resources.
Cost management
Better manage resource allocations for Azure Machine Learning Compute with workspace and resource level quota limits.
Only pay for what you need, with no upfront cost
For details, go to the Azure Machine Learning pricing page.
How to use Azure Machine Learning
Go to your studio web experience
Build and train
Deploy and manage
You can author new models and store your compute targets, models, deployments, metrics and run histories in the cloud.
Use automated machine learning to identify algorithms and hyperparameters, and track experiments in the cloud. You can also author models using notebooks or the drag-and-drop designer.
Deploy your machine learning model to the cloud or the edge, monitor performance and retrain it as needed.
Resources
Beginner tutorials
Advanced tutorials
Start using Azure Machine Learning today
Get instant access and a $200 credit by signing up for an Azure free account.
Sign in to the Azure portal.
Explore the documentation and tutorials. Find quickstarts and developer resources.
Customers using Azure Machine Learning
Jolie Vitale: Director of BI and Analytics, Carhartt"The model we deployed on Azure Machine Learning helped us choose the three new retail locations we opened in 2019. Those stores exceeded their revenue plans by over 200 percent in December, the height of our season, and within months of opening were among the best-performing stores in their districts."
Scandinavian Airlines
By using Azure Machine Learning, SAS is accurately identifying fraud with proficiency that wasn’t possible through manual methods. In the case of retroactively registering a flight for EuroBonus miles, a common source of fraud, the new system predicts fraud with 99 per cent accuracy.
Sze-Wan Ng: Director of Analytics & Development, Translink"With MLOps capabilities in Azure Machine Learning, we've improved bus departure predictions by 74 percent, and riders spend 50 percent less time waiting."
Dean Riddlesden, Senior Data Scientist, Global Analytics, Walgreens Boots Alliance"If I have 200 models to train—I can just do this all at once. It can be farmed out to a huge compute cluster, and it can be done in minutes. So I'm not waiting for days."
Alex Mohelsky: Partner and Advisory Data, Analytic and AI Leader, EY Canada"We see Azure Machine Learning and our partnership with Microsoft as critical to driving increased adoption and acceptance of AI from the regulators."
Xiaodong Wang: Chief Executive Officer, TalentCloud"The automated machine learning capabilities in Azure Machine Learning save our data scientists from doing a lot of time-consuming work, which reduces our time to build models from several weeks to a few hours."
Azure Machine Learning updates, blogs and announcements
-
ANNOUNCEMENT
Azure Open Datasets, now in preview, offers access to curated datasets.
-
UPDATE
Azure Machine Learning updates--November 2020
-
UPDATE
Azure Machine Learning offers added capabilities at lower cost
-
UPDATE
Azure Machine Learning updates Ignite 2020
-
UPDATE
Azure Machine Learning announces output dataset (Preview)
-
UPDATE
Azure Machine Learning studio web experience is generally available
Frequently asked questions about Azure Machine Learning
-
The service is generally available in several countries/regions, with more on the way.
-
The service-level agreement (SLA) for Azure Machine Learning is 99.9 per cent.
-
The Azure Machine Learning studio is the top-level resource for the machine learning service. It provides a centralised place for data scientists and developers to work with all the artefacts for building, training and deploying machine learning models.