Azure Machine Learning Service Accelerating machine learning workflows

Azure Machine Learning Service Accelerating machine learning workflows in the cloud Sharat Chikkerur Principal data science lead, Microsoft sharat@sharat. org Original slides by Wolfgang Pauli, Microsoft

Agenda • • Introduction to cloud services Azure Machine Learning Services • • • Overview Flavors: Iaa. S, Paa. S, Saa. S Demos • • • [Paa. S] Auto. ML [Saa. S] Azure Machine Learning Studio [Iaa. S/Paa. S] DSVM + Azure Machine Learning Service SDK

Flavors of Cloud Services Iaa. S Paa. S Saa. S (Infrastructure as a service) (Platform as a service) (Software as a service) • Provides self managed infrastructure as a service (compute, storage, networking). • Examples • Azure, AWS, GCP, Digital Ocean etc. • Provides managed services that can be used to build Saa. S solutions • Examples • App. Engine, Heroku, Azure. ML, Sage. Maker, Dataiku • Delivers full applications on the cloud • Fully managed service providing scalability & availability • Examples • Zoom, Dropbox, Azure. ML Studio, Stripe

Flavors of Machine learning platforms https: //www. gartner. com/doc/reprints? id=1 -25 D 1 CURH

Industrial ML: Machine learning workflow Build and train models Classical ML Data sources Deep learning Hyperparameter tuning Experimentation and pipelines Deployment Dev. Ops for data science

Industrial ML: AIOps/Dev. Ops Prepare Train & Test Model Register and Manage Model … 0101010010110100101101010 0100011110100101001010100101101001011010100100 011110100101000101110110100 10010010110101001000111101001010010100101101001 Build model (your favorite IDE) Prepare Data Deploy Service Monitor Model Build Image

Machine Learning on Azure Domain Specific Pretrained Models To reduce time to market Familiar Data Science Tools To simplify model development Popular Frameworks To build machine learning and deep learning solutions Productive Services To empower data science and development teams Powerful Hardware To accelerate deep learning … Vision Speech Language Search Py. Charm Jupyter Visual Studio Code Command line Tensor. Flow Scikit-Learn ONNX Py. Torch Azure Databricks Azure Machine Learning VMs CPU GPU FPGA From the Intelligent Cloud to the Intelligent Edge

Azure Machine Learning Service

Iaa. S: Data Science Virtual Machine

Data Science Virtual Machines (DSVM) Pre-Configured environments in the cloud for Data Science & AI Modeling, Development & Deployment. Samples to get started

Managed notebooks

Saa. S: Azure Machine Learning Designer (Studio)

Drag-n-drop machine learning workflows

Paa. S: Azure Machine Learning Service

Azure Machine Learning Service Set of Azure Cloud Services Python SDK That enables you to: ✓Prepare Data ✓Build Models ✓Train Models ✓Manage Models ✓Track Experiments ✓Deploy Models

Azure ML service Key Artifacts Workspace

Popular Frameworks Use your favorite machine learning frameworks Tensor. Flow Py. Torch Scikit-Learn without getting locked into one framework ONNX Community project created by Facebook and Microsoft Use the best tool for the job. Train in one framework and transfer to another for inference MXNet Chainer Keras

Training Infrastructure Dependencies and Containers Leverage system-managed AML compute or bring your own compute Distribute data Manage and share resources across a workspace Schedule jobs Train at cloud scale using a framework of choice Scale resources Autoscale resources to only pay while running a job Provision VM clusters Use the latest NDv 2 series VMs with the NVIDIA V 100 GPUs

Automated ML

Automated ML Algorithms Supported sklearn. linear_model. Logistic. Regression sklearn. linear_model. Elastic. Net sklearn. linear_model. SGDClassifier sklearn. ensemble. Gradient. Boosting. Regressor sklearn. naive_bayes. Bernoulli. NB sklearn. tree. Decision. Tree. Regressor sklearn. naive_bayes. Multinomial. NB sklearn. neighbors. KNeighbors. Regressor sklearn. svm. SVC sklearn. linear_model. Lasso. Lars sklearn. svm. Linear. SVC sklearn. linear_model. SGDRegressor sklearn. calibration. Calibrated. Classifier. CV sklearn. ensemble. Random. Forest. Regressor sklearn. neighbors. KNeighbors. Classifier sklearn. ensemble. Extra. Trees. Regressor sklearn. tree. Decision. Tree. Classifier lightgbm. LGBMRegressor sklearn. ensemble. Random. Forest. Classifier sklearn. ensemble. Extra. Trees. Classifier sklearn. ensemble. Gradient. Boosting. Classifier lightgbm. LGBMClassifier

Automated ML Use via the Python SDK

Experimentation Leverage service-side capture of run metrics, output logs and models Manage training jobs locally, scaled-up or scaled-out 80% 75% 90% 95% 85% Use leaderboards, side by side run comparison and model selection Conduct a hyperparameter search on traditional ML or DNN

Model Management Create/Retrain Model Register Model Monitor Enable Dev. Ops with full CI/CD integration with VSTS Track model versions with a central model registry Oversea deployments through Azure App. Insights

Deploy Azure ML models at scale Azure Machine Learning Service Cognitive Services Model Registry Your IDE Image Registry Cloud Service Azure Machine Learning Experimentation Heavy Edge Register Model External Model Scoring File Create & Register Image Light Edge Deployment & Model Monitoring

Deployments to Compute Targets

Summary

Azure Machine Learning service Bring AI to everyone with an end-to-end, scalable, trusted platform Boost your data science productivity Built with your needs in mind Automated machine learning Managed compute Increase your rate of experimentation Simple deployment Dev. Ops for machine learning Support for open source frameworks Deploy and manage your models everywhere Tool agnostic Python SDK Seamlessly integrated with the Azure Portfolio

Demos

Resources This talk https: //github. com/sharatsc/ub-cdse-2021 Azure Notebooks https: //github. com/Azure/Machine. Learning. Notebooks Azure ML Docs https: //docs. microsoft. com/en-us/azure/machine-learning/service/

Questions