Sign inSign up
Generative Models II thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Generative Models II

Published on Jul 27, 20177492 Views

Aaron Courville

DLSS & RLSS 2017 - Montreal

Related categories

Deep LearningReinforcement LearningUnsupervised Learning

Chapter list

Generative Models II00:00
Generative Modeling00:11
Generative Models II: Outline01:01
Autoregressive generative models - 102:07
Autoregressive generative models - 203:47
PixelCNN - 104:59
PixelCNN - 307:21
PixelCNN - 409:43
Improving PixelCNN10:49
Improving PixelCNN I11:30
PixelCNN: Experimental Results - 119:16
PixelCNN: Experimental Results - 219:36
PixelCNN: Experimental Results - 320:20
Improving PixelCNN II21:40
Parallel Multiscale Autoregressive Density Estimation - 123:03
Parallel Multiscale Autoregressive Density Estimation - 223:23
PixelCNN: Experimental Results - 423:50
PixelCNN - 225:07
Latent Variable Models - 126:23
Latent Variable Models - 227:36
Latent Variable Models - 331:50
Variational Auto-Encoder (VAE)32:13
VAE Inference model35:48
Reparametrization trick39:27
Vanilla VAE samples40:16
Training with backpropagation!41:33
Deep encoder/decoder: Some component collapse - 141:41
Deep encoder/decoder: Some component collapse - 242:21
PixelVAE - 145:37
PixelVAE Samples46:30
PixelVAE - 246:57
Inverse Autoregressive Flow49:14
Normalizing Flows - 150:29
Normalizing Flows - 252:00
VAE-IAF - 153:42
VAE-IAF - 255:14
Another way to train a latent variable model58:23
Generative Adversarial Networks - 159:03
Generative Adversarial Networks - 259:26
GAN Objective01:00:09
GAN Theory01:00:56
GAN Theory … in practice01:02:52
GAN samples01:04:30
Least-Squares GAN01:05:05
GAN Zoo01:05:53
An explo-GAN of papers01:06:16
DCGAN samples01:06:36
Cartoon of the Image manifold01:08:47
Training a GAN: Distances between Manifolds - 101:11:24
Training a GAN: Distances between Manifolds - 201:11:46
Jensen-Shannon Divergence - 101:12:09
What makes GANs special?01:13:16
Jensen-Shannon Divergence - 201:13:39
Earth-Movers Distance01:14:38
Wasserstein Distance - 101:15:38
Wasserstein Distance - 201:15:54
Wasserstein Distance - 301:16:06
Wasserstein GAN - 101:16:18
Wasserstein GAN - 201:18:25
Issues with Weight Clipping01:19:09
Gradient Penalty Approach - 101:20:10
Gradient Penalty Approach - 201:21:23
Comparison on difficult to train architectures - 101:21:52
Comparison on difficult to train architectures - 201:22:50
WGAN with Gradient Penalty01:23:12
But what about inference…01:24:22
ALI: model diagram01:24:32
Toy Example01:25:22
Hierarchical ALI - 201:26:45
Hierarchical ALI - 101:27:02