Labs Talk Podcast: AI on the Edge with Mike Tyka
Both AI and Edge Computing are hot topics, but the combination of both is even more fun, as they enable completely new use cases and customer experiences. Listen to our interview with Mike Tyka (Software Engineer, Google Research) if you want to find out how these trending topics come together or how this is related to TensorFlow and new cool products like Google Coral!
Mike Tyka is the creator of the (physcical) Teachable Machine. For the purpose of preparing for this podcast, I rebuilt the complete machine at home with my electronics knowledge and the help of my lasercutter for the bottom frame. If you like to see more about my remake of the teachable machine, have a look at this blog post.
The Teachable Machine uses an ML technique which is called transfer-learning to quickly learn / memorize new visual patterns. In the video below, I am training the machine for different products with the button clicks. Later on, when I hold the items into the field of view of the camera, the machine then is able to correctly identify the item again (you have to watch the LEDs closely). The processing speed of this Raspberry PI-powered machine is greatly enhanced due to the usage of a Google Coral TPU – a special Tensor Processing Unit which speeds up the processing of Tensorflow ML models.
- Guardian – Open AI too dangerous to release
- Open AI – Better Language Model
- KOI: Demo in a Minute
- SAP Now Tokyo
- Feature Extraction
- Digital Female Leader Award
- Green Keeper
- AI on the Edge
- Mike’s website
- Mike on Twitter
- Teachable Machine
- Google Coral & Teachable Machine