Save 37% off PRO during our Black Friday Sale! »

Predicting with GCP

2d2dbdf5d060b4c1bb238f8f59185cfb?s=47 Giulia
April 24, 2019

Predicting with GCP

20 minutes presentation for Women in Machine Learning & Data Science Paris meetup.

With the release of Google Cloud AutoML, Google Cloud Platform provides yet another out-of-the-box AI managed service. But this doesn’t mean that data scientists have no say in training and deploying customised machine learning models in the cloud. There are services, such as Cloud ML Engine, devoted to this specific goal.

2d2dbdf5d060b4c1bb238f8f59185cfb?s=128

Giulia

April 24, 2019
Tweet

Transcript

  1. @Giuliabianchl Predicting with Google Cloud Platform April 24, 2019 -

    Giulia Bianchi
  2. @Giuliabianchl @Giuliabianchl giulbia Data scientist > DXD-WiMLDS <

  3. @Giuliabianchl End of 2017 Google AutoML announcement GCP podcast 117

  4. @Giuliabianchl End of 2018 Google AI Hub announcement

  5. @Giuliabianchl @Giuliabianchl To access any of them you need to

    have access to GCP anyway...
  6. @Giuliabianchl • Set up a GCP account (gmail ID needed)

    • Sign in Google Cloud Console and set up a project ($$$) • Install Cloud SDK Google Cloud Platform
  7. @Giuliabianchl Did you say Cloud?

  8. @Giuliabianchl • Cloud → managed services • From "on-premises" to

    *aaS ◦ IaaS ◦ PaaS ◦ SaaS • Access to theoretically unlimited resources • Rapidity of provisioning • Reliability Cloud what ?! And why should I even care?
  9. @Giuliabianchl GCP AI services - non exhaustive! SaaS SaaS SaaS

    SaaS PaaS SaaS
  10. @Giuliabianchl @Giuliabianchl Cloud ML Engine… now known as AI Platform

  11. @Giuliabianchl AI Platform Doc

  12. @Giuliabianchl Parenting 2.0 giulbia/baby_cry_detection https://www.youtube.com/watch?v=N-LXrheCIKM

  13. @Giuliabianchl Training pipeline Input training data Feature engineering Train model

    SVM Model saved in laptop
  14. @Giuliabianchl Prediction pipeline Preliminary step Model deployed in RPi

  15. @Giuliabianchl Prediction pipeline Input new data Feature engineering Predict! Model

    deployed in RPi Prediction saved in RPi
  16. @Giuliabianchl Prediction pipeline Input new data Feature engineering Predict! Model

    deployed in RPi Prediction saved in RPi Done in 45 sec.
  17. @Giuliabianchl Parenting 3.0 giulbia/gcp-rpi giulbia/baby_cry_rpi

  18. @Giuliabianchl Prediction pipeline Preliminary step Model deployed in GCP

  19. @Giuliabianchl Prediction pipeline Input new data Feature engineering Predict! Model

    deployed in GCP Prediction saved in GCP
  20. @Giuliabianchl Prediction pipeline Input new data Feature engineering Predict! Model

    deployed in GCP Prediction saved in GCP Done in 5 sec.
  21. @Giuliabianchl @Giuliabianchl Some code

  22. @Giuliabianchl Cloud Function - main.py

  23. @Giuliabianchl Cloud Function - requirements.txt

  24. @Giuliabianchl Cloud Function - function deployment

  25. @Giuliabianchl Cloud Function - call to ML engine for prediction

  26. @Giuliabianchl Cloud Function - logs

  27. @Giuliabianchl @Giuliabianchl Take away

  28. @Giuliabianchl ... has a huge potential for data scientists ...

    is not fully data scientist friendly yet ... it needs more documentation ... evolves super fast Google Cloud Platform...
  29. @Giuliabianchl Thank you!

  30. JUNE, 27th 2019 - PAN PIPER, PARIS DATAXDAY.FR > DXD-WiMLDS

    <
  31. None
  32. None