Taskonomy: Disentangling Transfer Learning for Perception with Amir Zamir

TWiML Talk 164

In this episode I’m joined by Amir Zamir, Postdoctoral researcher at both Stanford & UC Berkeley.

Subscribe: iTunes / Google Play / Spotify / RSS

Amir joins us fresh off of winning the 2018 CVPR Best Paper Award for co-authoring “Taskonomy: Disentangling Task Transfer Learning.” In this work, Amir and his coauthors explore the relationships between different types of visual tasks and use this structure to better understand the types of transfer learning that will be most effective for each, resulting in what they call a “computational taxonomic map for task transfer learning.”

In our conversation, we discuss the nature and consequences of the relationships that Amir and his team discovered, and how they can be used to build more effective visual systems with machine learning. Along the way Amir provides a ton of great examples and explains the various tools his team has created to illustrate these concepts.


Be sure to sign up for our weekly newsletter. We recently shared a write up detailing the ML/AI Job Board we’re working on, and got a ton of encouragement and interest. To make sure you don’t miss anything, head over to twimlai.com/newsletter to sign up.

Vote for us!!

We’ve got some great news to share and also favor to ask! We’re in the running for this year’s People’s Choice podcast awards, in both the People’s Choice and Technology categories and we’d really appreciate your support! Head over to our nomination page to find out how to vote for us!!! Thanks in advance!

About Amir

Mentioned in the Interview

“More On That Later” by Lee Rosevere licensed under CC By 4.0

Originally published at twimlai.com on July 16, 2018.