Ready to use deep learning models

Ready to use deep learning models

7f378e07b7a5a685e7e273148d221a10?s=128

Gabriela de Queiroz

September 21, 2019
Tweet

Transcript

  1. R E A D Y T O U S E

    D E E P L E A R N I N G M O D E L S —
 Gabriela de Queiroz
 Sr. Engineering & Data Science Manager, IBM Founder, R-Ladies Founder, AI Inclusive @gdequeiroz | www.k-roz.com Slides: http://bit.ly/max-gdgdevfest19
  2. Center for Open Source Data and AI Technologies (CODAIT) -

    CODAIT aims to make AI solutions easier to create, deploy, and manage in the enterprise - Relaunch of the Spark Technology Center (STC) to reflect expanded mission Watson West Building 505 Howard St. San Francisco, California CODAIT codait.org @gdequeiroz | www.k-roz.com | gdq@ibm.com
  3. 30+ open source developers! • TensorFlow • PyTorch • Keras

    • Apache Arrow • ONNX • Jupyter • Apache Spark • Kubeflow • And more
  4. Machine Learning Team (10 open source developers) 1) TensorFlow 2)

    PyTorch 3) Keras 4) Apache Arrow
  5. Have you ever wanted to classify images, identify objects or

    generate captions of images? @gdequeiroz | www.k-roz.com | gdq@ibm.com Slides: http://bit.ly/max-gdgdevfest19
  6. With the Model Asset eXchange, you can!

  7. Applying Deep Learning: Perception Data ??? Train model ??? $$$

    Get model ??? Deploy model ??? $$$ Training – Data Scientist Consumption – App Developer, Domain Expert Slides: http://bit.ly/max-gdgdevfest19
  8. Applying Deep Learning: Reality Find model Get code Test, verify,

    fix Train model Deploy model Use model Discovery Execution Consumption 1 2 3 4A 4B 5 @gdequeiroz | www.k-roz.com | gdq@ibm.com
  9. Step 1: Find a model … that does what you

    need … that is free to use … that is performant enough
  10. Step 2: Get the code Is there a good implementation

    available? … that does what you need … that is free to use … that is performant enough TensorFlow code to build ResNet50 neural network graph
  11. Or … Step 2: Get the pre-trained weights Is there

    a good pre-trained model available? … that does what you need … that is free to use … that is performant enough
  12. Step 3: Verify the model you found Check … …

    that does what you need … that is free to use (license) … that is performant enough (computation & accuracy)
  13. Step 4 (a): Train the model

  14. Step 4 (a): Train the model

  15. Step 4 (b): Figure out how to deploy your model

    … adjust inference code (or write from scratch) … package inference code and model code, and pre-trained weights together … deploy your package
  16. Step 5: Consume your model … plug in into your

    application
  17. Step 6: Profit … hopefully

  18. Applying Deep Learning: Reality Find model Get code Test, verify,

    fix Train model Deploy model Use model Discovery Execution Consumption 1 2 3
  19. Model Asset eXchange • One-stop place for developers/data scientists to

    find and use free and open source deep learning models ibm.biz/model-exchange @gdequeiroz | www.k-roz.com | gdq@ibm.com
  20. Model Asset eXchange (MAX) • Wide variety of domains (text,

    audio, image, etc) • Multiple deep learning frameworks • Vetted and tested code and IP • Trainable and Deployable versions • Build and deploy a model web service in seconds ibm.biz/model-exchange
  21. What do I need to get started?

  22. None
  23. Find* a state-of-art open source deep learning model specific to

    domain Validate license terms Perform model health check & code clean up Wrap models in MAX framework and provide REST API Publish the deployable model as Docker images on Docker Hub Use the MAX training framework to create an image for custom model training Review and Continuous Integration * or build from scratch Model Asset eXchange (MAX) ibm.biz/model-exchange BEHIND THE SCENES
  24. Ways of accessing the model

  25. OBJECT DETECTOR Localize and identify multiple objects in a single

    image @gdequeiroz | www.k-roz.com | gdq@ibm.com
  26. None
  27. None
  28. 1) REST API MODEL REST API flask Request information from

    model Send input to model GET POST
  29. 1.1) Via Swagger

  30. 1.2) Via Python You can try! http://ibm.biz/max-notebook

  31. 1.2) Via R

  32. 2) WEB APP • User uses Web UI to send

    an image to Model API • Model API returns object data and Web UI displays detected objects •
  33. 2) WEB APP bit.ly/object-detector-app

  34. 3) Node-RED flow

  35. 4) CodePen

  36. What if I want to train a model?

  37. None
  38. None
  39. And there is more!

  40. MAX-Framework MAX-Skeleton •A pip installable python library •Wrapper around flask

    •Abstracts out all basic functionality of the MAX model into MAXModelWrapper and MAXApi abstract classes github.com/IBM/MAX-Framework •Template to create a deployable MAX model •Contains all the code scaffolding and imports MAX Framework github.com/IBM/MAX-Skeleton
  41. All that is available for YOU for FREE @gdequeiroz |

    www.k-roz.com | gdq@ibm.com
  42. How do I get started? @gdequeiroz | www.k-roz.com | gdq@ibm.com

  43. @gdequeiroz | www.k-roz.com | gdq@ibm.com bit.ly/max-tutorial

  44. @gdequeiroz | www.k-roz.com | gdq@ibm.com Code Patterns How to easily

    consume MAX models bit.ly/max-code-patterns
  45. @gdequeiroz | www.k-roz.com | gdq@ibm.com

  46. Data Asset eXchange (DAX) • Curated free and open datasets

    under open data licenses • Standardized dataset formats and metadata • Ready for use in enterprise AI applications • Complement to the Model Asset eXchange (MAX) • ibm.biz/data-asset-exchange
  47. Data Asset eXchange (DAX)

  48. Data Asset eXchange (DAX)

  49. Thank you! K- ROZ .COM @GDEQUEIROZ GABRI EL A DE

    QU EI ROZ ai-inclusive.org Our mission is to increase the representation and participation of gender minority groups in AI. info@ai-inclusive.org ibm.biz/model-exchange Slides: http://bit.ly/max-gdgdevfest19