Interpreting Visual Concepts
Ishan Misra discusses the interpretability of the twelve x twelve grid output in visual concepts, explaining how it can be used to localize objects in an image. He also touches on the connection between psychology and machine learning, specifically the prototypical properties of objects. The conversation concludes with a discussion on the trend towards using larger web-scale datasets for research in this area.In this clip
From this podcast
Data Skeptic
Machine Learning on Images with Noisy Human-centric Labels
Related Questions