ICLR 2020: Yann LeCun and Energy-Based Models
Machine Learning Street Talk (MLST)
00:00
Energy-Based Models in Machine Learning
This chapter investigates energy-based models and their applications in machine learning, emphasizing their ability to optimize labels through differentiable representations. It covers challenges in deep learning, including the need for fewer labeled samples and reasoning, while connecting energy functions to practical examples such as GANs and k-means clustering. The discussion also touches upon recent developments from the ICLR 2020 conference, highlighting key insights and ongoing debates within the realm of machine learning.
Transcript
Play full episode
