Machine Learning Street Talk (MLST) cover image

#55 Self-Supervised Vision Models (Dr. Ishan Misra - FAIR).

Machine Learning Street Talk (MLST)

00:00

The Challenge of Naming and Abstractions in Machine Learning

Naming things is a tricky problem, with different languages having different words for the same thing. Understanding what a model has learned is difficult because the concept of the right abstraction varies from person to person and culture to culture. When we try to interpret things, we inject our own bias into the model. For example, should a cartoon banana and a real banana be considered the same? This raises a fascinating philosophical question about our abstract concepts.

Play episode from 36:49
Transcript

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app