[Machine Learning 15minutes! Broadcast #67] Azure AI - Build 2022 Updates and more...

https://satonaoki.wordpress.com/2022/06/26/ml15min-azure-ai-build-2022-updates/

Transcript

1. Azure AI – Build 2022 Updates and more...
   SATO Naoki (Neo)
   Senior Software Engineer, Microsoft
   

   View Slide

2. View Slide

3. How customers adopt AI
    Knowledge
   Mining
   60
   %
    Reduction in cost
   of compliance
    Conversational
   AI
   30
   %
    Of all e-commerce will be
   through voice by 2030
    Document Process
   Automation
   46
   %
    Cost savings with process
   automation in IT enterprises
    Machine
   Translation
   78
   %
    Of language services utilize
   Machine Translation
    Speech
   Transcription &
   Analytics
   64%
    Of online education
   experts use automated
   transcription
   Build, deploy and manage custom ML models
   Solve common business scenarios
   Knowledge
   Mining
   Conversational
   AI
   Business Process
   Automation
   Machine
   Translation
   Speech Transcription
   & Analytics
   

   View Slide

4. View Slide

5. ML platform
   Azure Machine Learning
   Azure AI
   Vision Speech Language Decision OpenAI Service
   Customizable, high quality AI models
   Cognitive Services
   Bot Service Cognitive Search Form Recognizer Video Indexer Metrics Advisor Immersive Reader
   Scenario specific AI services
   Applied AI Services
   

   View Slide

6. Preview
   Preview
   Preview
   Auto-tuning capability Preview
   Gated Preview
   Gated Preview
   Gated Preview
   

   View Slide

7. Preview
   Document & Conversation Summarization Preview
   Custom Named Entity Recognition Generally Available
   Custom Text Classification Generally Available
   Conversational Language Understanding Generally Available
   Document Translation Generally Available
   

   View Slide

8. Public Preview
   

   View Slide

9. GitHub Copilot
   https://github.com/features/copilot/
   

   View Slide

10. ML platform
    Azure Machine Learning
    Azure AI
    Vision Speech Language Decision OpenAI Service
    Customizable, high quality AI models
    Cognitive Services
    Bot Service Cognitive Search Form Recognizer Video Indexer Metrics Advisor Immersive Reader
    Scenario specific AI services
    Applied AI Services
    

    View Slide

11. Data
    Preparation
    Model
    Building
    Model
    Training
    Model
    Management &
    Monitoring
    Model
    Deployment
    Responsible AI
    MLOps
    

    View Slide

12. Structured Data
    Unstructured Data Analytics
    Business Apps
    Prepare Data Build & Train Manage & Monitor
    Deploy
    across clouds and on-premises
    Serverless
    Compute
    Managed
    Kubernetes
    Azure Edge & Hybrid
    Azure Arc-enabled
    Kubernetes
    Edge/IoT Devices
    Azure ML
    

    View Slide

13. MLOps
    

    View Slide

14. Preview
    Generally Available
    Preview
    Generally Available
     Support for natural language processing and image tasks
     Generation of model training codes
     Enhancements for product integration and MLOps
    

    View Slide

15. Azure ML
    Azure ML
    Your editor of choice
    

    View Slide

16. Pipeline
    Metadata
    name, display_name, version, type, etc.
    Interface
    input/output specifications (name, type,
    description, default value, etc).
    Command, Code & Environment
    command, code and environment required
    to run the component
    SDK
    CLI
    UI
    

    View Slide

17. Fairness Inclusiveness
    Privacy &
    Security
    Transparency
    Reliability &
    Safety
    Accountability
    Microsoft’s AI
    Principles
    

    View Slide

18. Responsible AI investments and safeguards for facial recognition
    https://azure.microsoft.com/en-us/blog/responsible-ai-investments-and-safeguards-for-facial-recognition/
    

    View Slide

19. View Slide

20. Azure Machine Learning
    Responsible AI Dashboard
    & Responsible AI Scorecard
    Managed endpoints
    Python SDK v2
    Arc support
    Command Line Interface v2
    Custom reusable
    components (in pipelines)
    Automated ML new features
    Support for natural language
    processing and image tasks
    Generation of model
    training codes
    Enhancements for product
    integration and MLOps
    Integration with
    RStudio Workbench
    

    View Slide

21. View Slide