[13] Adji Bousso Dieng - Deep Probabilistic Graphical Modeling
The Thesis Review
00:00
Reweighted EM and Entropy Regularized Adversarial Learning
The VAE is entangling all of that into one, into one step and doing updates simultaneously for both the model parameters and the variation of parameters. The reweighted wakeslip was presented as an important sampling extension of the wake slip algorithm. And the highway was presented as a way to optimize a title low bound than the elbow. But really, it's also an important sampling Extension of the VAU objective Yeah. Going from AM to this deep learning setting, we ended up with algorithms that in some ways are more generalizable.
Transcript
Play full episode
