AI on the Edge

I presented this deck at Global AI Nights Istanbul. The deck is about running containerized AI logic on edge devices through Azure IoT Hub and IoT Edge Runtime.

Transcript

1. Daron Yöndem
   http://daron.me
   @daronyondem
   

   View Slide

2. An edge that’s intelligent and cloud-enabled.
   That’s the future as we see it.
   

   View Slide

3. Cloud : Globally available, unlimited compute resources
   IoT : Harnessing signals from sensors and devices, managed
   centrally by the cloud
   Edge : Intelligence offloaded from the cloud to IoT devices
   AI : Breakthrough intelligence capabilities
   

   View Slide

4. Azure IoT
   Hub
   Azure IoT
   Suite
   Microsoft IoT
   Central
   Windows 10
   IoT Core
   Supports the
   languages and
   frameworks you
   already know
   Management
   NUI
   Edge compute
   Fully managed
   and hosted by
   Microsoft
   Coming soon
   Remote
   monitoring
   Predictive
   maintenance
   Connected
   factory
   Secure, scalable
   PaaS
   For connecting
   any device
   running any OS
   

   View Slide

5. Cold Path: Azure Machine Learning
   • Build powerful, cloud-based machine learning applications
   • Quickly create and deploy analytics models with ready-to-use algorithms libraries
   • Includes hundreds of R and Python build-in packages in addition to supporting custom code.
   Hot Path: Azure Stream Analytics
   • Analyze real-time and on demand data to power intelligent actions
   • Easy to set up with SQL-like language
   • Connects directly to Azure IoT Hub for stream ingestion and hot path analysis
   Warm Path: Azure Time Series Insights
   • Visualize IoT time series data in near real time
   • Easy to get started with no up-front data modeling required
   • Customize your own solution via REST query APIs
   

   View Slide

6. Straightforward
   Business Logic
   Smart Building
   Protocol
   Translation
   Industrial
   IoT Adapters
   Autonomous
   Driving
   Straightforward
   Filtering
   Batch Data
   Processing
   Home Automation
   Hubs
   Smart
   Meeting Solutions
   w/ People Recognition
   Custom Code Requirements
   Reliability Requirements
   Low Medium High
   Low
   Portable Modules / Functions
   Medium
   E.g.: Custom Code in Containers
   High
   E.g.: ML/DNNs in Containers
   

   View Slide

7. IoT Hub
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Device
   Twin
   Jobs
   Schedule and Broadcast
   Device Twin Changes
   Set Desired Properties,
   Tags, Call Methods
   Queries
   Query Across
   Device Twin State
   For Business Logic,
   Reporting and
   Compliance
   

   View Slide

8. • Container based modules
   • Azure Functions
   • Azure Stream Analytics
   • Azure Machine Learning
   • Cognitive Services
   • Offline / Synchronized Device Twins
   • Local Storage
   • Cloud Management & Deployment
   

   View Slide

9. Azure IoT Edge IoT Hub
   Devices
   Local Storage
   Azure Machine
   Learning
   (Container)
   Functions
   Runtime
   Container
   Management
   Device
   Twin
   Device
   Twin
   Azure Stream
   Analytics
   (Container)
   Azure Functions
   (Container)
   Cognitive
   Services
   (Container)
   Custom Code
   (Container)
   Module
   Twin
   Module
   Twin
   Module
   Twin
   Module
   Twin
   Module
   Twin
   Module
   Twin
   Module
   Twin
   Module
   Twin
   Module
   Twin
   Module
   Twin
   

   View Slide

10. Running ML on Azure Functions on the Edge
    

    View Slide

11. View Slide

12. View Slide

13. View Slide

14. View Slide

15. View Slide

16. View Slide

17. View Slide

18. View Slide

19. IoT Edge Runtime
    https://drn.fyi/2k1otsV
    Azure IoT Tools to VS Code
    https://drn.fyi/2kw6Mlj
    Azure Functions IoT Edge Deployment
    https://drn.fyi/2m0Eksm
    

    View Slide

20. http://daron.me | @daronyondem
    Download slides here;
    http://decks.daron.me
    

    View Slide