Machine Learning Street Talk (MLST) cover image

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

Machine Learning Street Talk (MLST)

00:00

The Nuances of Neural Networks and Conceptualization

This chapter explores the intricate relationship between neural networks and their ability to generalize concepts, using the metaphor of bananas in various representations. It critically examines the challenges in object recognition, the impact of biases, and the philosophical implications surrounding common sense knowledge in AI. The conversation also addresses the limitations of self-supervised learning models in achieving true semantics and the necessity for a reevaluation of evaluation metrics in vision systems.

Transcript
Play full episode

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