Disentangled Representations
Olivier shares his journey from economics to machine learning and discusses his work on disentangled representations at Google Brain. He explains how these representations can capture distinct factors of variation in high-dimensional data, such as images. Through a large-scale study of over 10,000 models, surprising insights emerged regarding the effectiveness of various approaches in unsupervised learning.In this clip
From this podcast
The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Disentangled Representations & Google Research Football with Olivier Bachem - TWIML Talk #293
Related Questions