Dec19 Meetup: ML & AI on Azure by Marc Schöni

Dec19 Meetup: ML & AI on Azure by Marc Schöni

Learn about the solution components available through the Azure cloud for building cutting edge ML & AI solutions at scale.

About Marc Schöni:
Marc is a Cloud Solution Architect at Microsoft with a wealth of experience for data science solutions across industries. On a daily basis he works with customers on designing real-world Azure based ML & AI solutions. Marc holds a Master’s degree in Data Science and is a lecturer on ML at University of St. Gallen.

You can find him at:
https://www.linkedin.com/in/aiguy

0754d30f3acc99a940aebdcd49d5af97?s=128

Azure Zurich User Group

December 17, 2019
Tweet

Transcript

  1. 5.

    Language Vision Speech Decision Web search Conversation transcription capability Language

    Understanding Ink Recognizer Personalizer Computer Vision Neural Text-to-Speech Anomaly Detector Speech Service Device SDK QnA Maker Most comprehensive list of cognitive capabilities
  2. 6.

    The most comprehensive pre-trained AI Language Vision Speech Decision Web

    search Bing Spell Check Custom Vision Personalizer Form Recognizer Neural Text-to-Speech Anomaly Detector Content Moderator Content Moderator Custom Speech Speech transcription Text-to-Speech Conversation transcription capability Face Video Indexer Ink Recognizer Computer Vision Language Understanding QnA Maker Text Analytics Translator Text Bing Web Search Bing Custom Search Bing Video Search Bing Image Search Bing Local Business Search Bing Visual Search Bing Entity Search Bing News Search Bing Autosuggest
  3. 8.

    4K Community commits 2M+ SDK downloads Microsoft Bot Framework Easily

    deploy and manage Build with open source tools Azure Bot Service 35K Active bots 3K Bots created weekly
  4. 13.
  5. 15.

    Ingest Extract experience 2 3 1 document primaryImage secondaryImage job

    company duration job company duration job company duration
  6. 16.

    Cognitive search capability Ingest Candidates with 5 years experience Extract

    Search experience 2 3 1 document primaryImage secondaryImage job company duration job company duration job company duration
  7. 19.

    Quickly extract text, tables and key value pairs with simple

    REST API calls Start with any type of forms Preview W-2 Receipt New
  8. 23.

    Domain specific pretrained models To simplify solution development Popular frameworks

    To build advanced deep learning solutions Productive services To empower data science and development teams Powerful infrastructure To accelerate deep learning Familiar data science tools To simplify model development From the Intelligent Cloud to the Intelligent Edge Azure Databricks Machine Learning VMs TensorFlow PyTorch ONNX Language Speech … Decision Vision Scikit-Learn Azure Notebooks Jupyter Visual Studio Code Command line Azure Machine Learning CPU GPU FPGA Web search
  9. 25.

    Azure Machine Learning Any tool + any framework Automated ML

    + drag & drop + code first For all skill levels Open Integrated with Azure DevOps Industry leading MLOps
  10. 26.

    Automated machine learning UI Visual interface Machine learning notebooks New

    capabilities in Azure Machine Learning service Centralized model registry
  11. 29.

    Mileage Condition Car brand Year of make Regulations … Parameter

    1 Parameter 2 Parameter 3 Parameter 4 … Gradient Boosted Nearest Neighbors SVM Bayesian Regression LGBM … Mileage Gradient Boosted Criterion Loss Min Samples Split Min Samples Leaf Others Model Which algorithm? Which parameters? Which features? Car brand Year of make
  12. 30.

    Which algorithm? Which parameters? Which features? Mileage Condition Car brand

    Year of make Regulations … Gradient Boosted Nearest Neighbors SGD Bayesian Regression LGBM … Nearest Neighbors Criterion Loss Min Samples Split Min Samples Leaf XYZ Model Iterate Gradient Boosted N Neighbors Weights Metric P ZYX Mileage Car brand Year of make Car brand Year of make Condition Track
  13. 32.

    Enter data Define goals Apply constraints Output Input Intelligently test

    multiple models in parallel Optimized model 95%
  14. 33.

    Feature Importance Mileage Condition Car brand Year of make Regulations

    0 1 Feature Importance Mileage 0 1 Condition Car brand Year of make Regulations 95% model 70% model Enable model explain-ability for every automated ML iteration, not just the optimal model
  15. 35.

    Prepare data Build & train models Deploy & predict Data

    storage locations Data ingestion DATA PREPARATION MODEL BUILDING & TRAINING MODEL DEPLOYMENT Normalization Transformation Validation Featurization Hyper-parameter tuning Automatic model selection Model testing Model validation Deployment Batch scoring
  16. 36.

    Prepare data Build & train models Deploy & predict Data

    storage locations Data ingestion DATA PREPARATION MODEL BUILDING & TRAINING MODEL DEPLOYMENT Normalization Transformation Validation Featurization Hyper-parameter tuning Automatic model selection Model testing Model validation Deployment Batch scoring Normalization Transformation Validation Featurization Hyper-parameter tuning Automatic model selection Model testing Testing error
  17. 37.

    Prepare data Build & train models Deploy & predict Data

    storage locations Data ingestion DATA PREPARATION MODEL BUILDING & TRAINING MODEL DEPLOYMENT Normalization Transformation Validation Featurization Hyper-parameter tuning Automatic model selection Model testing Model validation Deployment Batch scoring Normalization Transformation Validation Featurization Hyper-parameter tuning Automatic model selection Model testing Model testing Model validation Deployment Batch scoring Error resolved Testing error
  18. 38.

    Prepare data Build & train models Deploy & predict Data

    storage locations Data ingestion DATA PREPARATION MODEL BUILDING & TRAINING MODEL DEPLOYMENT Normalization Transformation Validation Featurization Hyper-parameter tuning Automatic model selection Model testing Model validation Deployment Batch scoring Normalization Transformation Validation Featurization Hyper-parameter tuning Automatic model selection Model testing Model validation New data Deployment Batch scoring
  19. 39.

    Unattended runs Schedule a few steps to run in parallel

    or in sequence to focus on other tasks while your pipeline runs Mixed and diverse compute Use multiple pipelines that are reliably coordinated across heterogeneous and scalable computes and storages Reusability Create templates of pipelines for specific scenarios such as retraining and batch scoring Tracking and versioning Name and version your data sources, inputs and outputs with the pipelines SDK
  20. 41.

    Build advanced deep learning solutions Use your favorite deep learning

    frameworks 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 TensorFlow PyTorch Scikit-Learn MXNet Chainer Keras
  21. 42.

    Simplify deployment + accelerate inferencing with ONNX runtime Track models

    in production Capture model telemetry From the Intelligent Cloud to the Intelligent Edge On-premises Cloud