Azure Machine Learning Studio

Documentation

1. Azure Machine Learning Documentation
2. Overview
    2.1. What is Azure Machine Learning?
    2.2. Azure Machine Learning vs Studio (classic)
    2.3. Architecture & terms
3. Tutorials
    3.1. Studio
        3.1.1. Designer (drag-n-drop)
            3.1.1.1. 1. Train a regression model
            3.1.1.2. 2. Deploy that model
        3.1.2. Automated ML (UI)
            3.1.2.1. Create automated ML experiments
            3.1.2.2. Forecast demand (Bike share data)
        3.1.3. Label image data
    3.2. Python SDK
        3.2.1. Create first ML experiment
            3.2.1.1. 1. Set up workspace & dev environment
            3.2.1.2. 2. Train your first model
        3.2.2. Image classification (MNIST data)
            3.2.2.1. 1. Train a model
            3.2.2.2. 2. Deploy a model
        3.2.3. Regression with Automated ML (NYC Taxi data)
            3.2.3.1. Auto-train an ML model
        3.2.4. Batch score models with pipelines
        3.2.5. Prep your code for production
    3.3. R SDK
        3.3.1. Create first ML experiment (R)
    3.4. Machine Learning CLI
        3.4.1. Train & deploy with CLI
        3.4.2. Create workspaces with ARM template
    3.5. Visual Studio Code
        3.5.1. Set up Azure Machine Learning extension
        3.5.2. Train and deploy a TensorFlow image classification model
4. Samples
    4.1. Jupyter Notebooks
    4.2. Designer examples & datasets
    4.3. End-to-end MLOps examples
    4.4. Open Datasets (public)
5. Concepts
    5.1. Enterprise vs. Basic editions
    5.2. Plan and manage costs
    5.3. Designer (drag-n-drop ML)
        5.3.1. Designer overview
        5.3.2. Algorithm cheat sheet
        5.3.3. How to select algorithms
    5.4. Automated ML
        5.4.1. Automated ML overview
        5.4.2. Overfitting & imbalanced data
    5.5. Workspace
    5.6. Environments
    5.7. Compute instance
    5.8. Compute target
    5.9. Data
        5.9.1. Data access
        5.9.2. Data ingestion
        5.9.3. Data processing
    5.10. Model training
    5.11. Distributed training
    5.12. Deep learning
    5.13. Model management (MLOps)
    5.14. Model portability (ONNX)
    5.15. ML pipelines
    5.16. Enterprise readiness & security
        5.16.1. Enterprise security and governance
            5.16.1.1. Enterprise security overview
            5.16.1.2. Manage users and roles
            5.16.1.3. Use virtual networks
            5.16.1.4. Use Private Link
            5.16.1.5. Secure web services with TLS
            5.16.1.6. Use Azure AD identity in AKS deployments
            5.16.1.7. Regenerate storage access keys
            5.16.1.8. Use Azure Firewall
            5.16.1.9. Secure coding
        5.16.2. Set up authentication
        5.16.3. Monitor Azure Machine Learning
    5.17. Responsible ML
        5.17.1. Responsible ML overview
        5.17.2. Model interpretability
        5.17.3. Fairness in Machine Learning
        5.17.4. Differential Privacy
6. How-to guides
    6.1. Create & manage workspaces
        6.1.1. Use Azure portal
        6.1.2. Use Azure CLI
        6.1.3. Use REST
        6.1.4. Use Resource Manager template
    6.2. Set up your environment
        6.2.1. Set up dev environments
        6.2.2. Run Jupyter Notebooks
        6.2.3. Set up software environments
        6.2.4. Enable logging
        6.2.5. Set input & output directories
        6.2.6. Interactive debugging
        6.2.7. Git integration
    6.3. Work with data
        6.3.1. Label data
            6.3.1.1. Get data labeled
            6.3.1.2. Label images
            6.3.1.3. Create datasets with labels
        6.3.2. Get data
            6.3.2.1. Data ingestion with Azure Data Factory
            6.3.2.2. DevOps for data ingestion
            6.3.2.3. Import data in the designer
        6.3.3. Access data
            6.3.3.1. Connect to Azure Storage
            6.3.3.2. Get data from a datastore
        6.3.4. Manage & consume data
            6.3.4.1. Train with datasets
            6.3.4.2. Detect drift on datasets
            6.3.4.3. Version & track datasets
            6.3.4.4. Preserve data privacy
    6.4. Train models
        6.4.1. Set up training environments
        6.4.2. Use estimators for ML
            6.4.2.1. Create estimators in training
            6.4.2.2. Tune hyperparameters
        6.4.3. Scikit-learn
        6.4.4. TensorFlow
        6.4.5. Keras
        6.4.6. PyTorch
        6.4.7. Track & monitor training
            6.4.7.1. Start, monitor or cancel runs
            6.4.7.2. Log metrics for training runs
            6.4.7.3. Track experiments with MLflow
            6.4.7.4. Visualize runs with TensorBoard
        6.4.8. Interpret & explain models
            6.4.8.1. Interpret ML models
            6.4.8.2. Explain automated ML models
        6.4.9. Assess and mitigate model fairness
        6.4.10. Use Key Vault when training
        6.4.11. Reinforcement learning
    6.5. Automated machine learning
        6.5.1. Use automated ML (Python)
        6.5.2. Use automated ML (interface)
        6.5.3. Use remote compute targets
        6.5.4. Auto-train a forecast model
        6.5.5. Data splits & cross-validation (Python)
        6.5.6. Featurization in automated ML (Python)
        6.5.7. Use automated ML in ML pipelines (Python)
        6.5.8. Understand charts and metrics
    6.6. Deploy & serve models
        6.6.1. Where and how to deploy
        6.6.2. Deployment scenarios
            6.6.2.1. Azure ML compute instances
            6.6.2.2. Azure Kubernetes Service
            6.6.2.3. Azure Container Instances
            6.6.2.4. GPU inference
            6.6.2.5. Azure App Service
            6.6.2.6. Azure Functions
            6.6.2.7. Azure Cognitive Search
            6.6.2.8. Azure IoT Edge devices
            6.6.2.9. FPGA inference
            6.6.2.10. Custom Docker images
            6.6.2.11. Non-Azure ML models
        6.6.3. Troubleshoot & debug
        6.6.4. Call service endpoint
        6.6.5. Monitor models
            6.6.5.1. Collect & evaluate model data
            6.6.5.2. Detect data drift
            6.6.5.3. Monitor with Application Insights
        6.6.6. Deploy encrypted inferencing service
    6.7. Build & use ML pipelines
        6.7.1. Create ML pipelines (Python)
        6.7.2. Moving data into and between ML pipeline steps (Python)
        6.7.3. Schedule a pipeline (Python)
        6.7.4. Trigger a pipeline
        6.7.5. Debug & troubleshoot pipelines
        6.7.6. Debug pipelines in Application Insights
        6.7.7. Designer transform data
        6.7.8. Designer retrain using published pipelines
        6.7.9. Designer batch predictions
        6.7.10. Designer execute Python code
        6.7.11. Use parallel run step
        6.7.12. Debug & troubleshoot parallel run step
        6.7.13. Iterating and evolving machine learning pipelines
    6.8. Azure Pipelines for CI/CD
    6.9. Manage resource quotas
    6.10. Manage resources VS Code
    6.11. Export and delete data
    6.12. Create event driven workflows
7. Reference
    7.1. Python SDK
    7.2. R SDK
    7.3. CLI
    7.4. REST API
    7.5. Designer module reference
    7.6. Monitor data reference
    7.7. Machine learning pipeline YAML reference
8. Resources
    8.1. Release notes
    8.2. Azure roadmap
    8.3. Pricing
    8.4. Regional availability
    8.5. Known issues
    8.6. User forum
    8.7. Stack Overflow
    8.8. Compare our ML products
    8.9. What happened to Workbench
    8.10. Designer accessibility features
    8.11. Curated Environments