Stanford CS236: Deep Generative Models I 2023 I Lecture 9 - Normalizing Flows

For more information about Stanford's Artificial Intelligence programs visit: stanford.io/ai
To follow along with the course, visit the course website:
deepgenerativemodels.github.io/
Stefano Ermon
Associate Professor of Computer Science, Stanford University
cs.stanford.edu/~ermon/
Learn more about the online course and how to enroll: online.stanford.edu/courses/c...
To view all online courses and programs offered by Stanford, visit: online.stanford.edu/

[@user-zr4ns3hu6y]

I think the titles of lecture 8 and lecture 9 have been switched.

[@CPTSMONSTER]

15:15 High likelihood and bad samples, garbage component is a constant in log-likelihood
40:00? Expectation on p data and p theta, how was this chosen
46:35 Note optimization of phi (discriminator) and theta (generator of fake samples)
50:45 Likelihood model in discriminator, but GANs can avoid likelihoods
1:00:15? Expectation on p data and p theta, added?
1:06:50 Minimax training objective
1:15:00 GANs no longer state of the art, very hard to train, mode collapse, no clean loss function to evaluate