Myths and Reality of Android Machine Learning

Tuesday 1:30 pm, Bagan Ballroom

About This Talk

Machine Learning (ML) in the Cloud is established. On-device ML is just getting started. In this talk, I will talk about the role of Android and cover a broad range of topics including TensorFlow, ML Kit and Android framework.

Even among machine learning experts, we have encountered many myths about on-device machine learning:

  1. Mobile devices are not powerful enough for machine learning.
  2. Machine learning is not for me, it’s for people with PhDs.
  3. Machine learning requires a lot of data - millions and billions of samples.

In this talk, Hoi will bust some myths with live demos and introduce best practices for developers to get around the reality of building everyday ML. Use cases including new in-person experience using the object detection and tracking. Also, the Android speech related APIs might surprise you… in a good way.

Photo of

Hoi Lam

Hoi is the Head of Developer Relations for Android Machine Learning and also a sector executive for the Institution of Engineering and Technology (IET). He was part of the launch team for Wear OS by Google (previously, Android Wear), Chromecast, Google Glass Enterprise Edition, and the Google Beacon Platform. Prior to Google, he founded Exahive to disrupt the mobile commerce market place. Hoi was an Equity Research VP at Deutsche Bank and Citigroup covering technology companies. Before banking, he was a product manager / application architect at Accenture. Hoi holds a Master and Bachelor degree in Space Engineering from the University of Cambridge.

Android is a trademark of Google LLC. The Android, Google Play, Google Play logo, and
Google Inc.'s logo are registered trademarks of Google Inc.

© 2019 DroidYangon