The State Of Machine Learning in iOS 13

The State Of Machine Learning in iOS 13

Apple's Machine Learning ecosystem is getting more and more powerful every year. This talk:

- takes a look back to Apple's Machine Learning history all the way to the modern days
- highlights best uses of Machine Learning within iOS, Apple apps, and third party apps
- helps developers getting started adapting Machine Learning in their own apps

30 minutes talk.

E35f18d9e5584b48a7cf5550f905522f?s=128

Federico Zanetello

November 19, 2019
Tweet

Transcript

  1. The State Of Machine Learning in iOS 13 Federico Zanetello

    ★★★★★ fivestars.blog • @zntfdr
  2. None
  3. Why care?

  4. None
  5. WWDC 2017 4 videos, ~150 minutes WWDC 2018 7 videos,

    ~250 minutes WWDC 2019 15 videos, ~400 minutes
  6. None
  7. None
  8. None
  9. None
  10. AI ➡ Software ➡ World

  11. None
  12. Vision Face detection Face features detection Text, bar code, shape

    detection Object tracking Image classification Object detection Attention image saliency Objectness image saliency Image similarity
  13. Natural Language Sentiment Analysis Word Tagging Word Embedding Text Classification

    Translation
  14. Speech & SoundAnalysis Speech recognition (live or prerecorded) Transcriptions Alternative

    interpretations Over 50 languages supported Jitter, shimmer analysis Pitch, voicing analysis Sound classification
  15. None
  16. None
  17. None
  18. None
  19. None
  20. None
  21. None
  22. None
  23. None
  24. None
  25. On-device Training

  26. Examples? FaceID iOS Keyboard & QuickType Photos.app Hey Siri &

    Siri Watch Maps.app Store.app Safari.app ARKit
  27. None
  28. Credits & Resources developer.apple.com/machine-learning developer.apple.com/documentation/createml github.com/apple/turicreate github.com/apple/coremltools developer.apple.com/design/human-interface-guidelines/machine-learning techinsights.com developer.apple.com/videos/all-videos

  29. The State Of Machine Learning in iOS 13 Federico Zanetello

    ★★★★★ fivestars.blog • @zntfdr