In this lecture, Chin-Wei will talk about a form of autoencoder known as the Variational Autoencoder (VAE). We’ll see how a deep latent gaussian model can be seen as an autoencoder via Amortized variational inference, and how such an autoencoder can be used as a generative model. At the end, we’ll take a look at variants of VAE and different ways to improve inference.
- Variational Autoencoders (updated) by CW Huang
Reference: (* = you are responsible for this material)
- *Sections 20.10.3 of the Deep Learning textbook.
- Auto-Encoding Variational Bayes by Diederik Kingma (ICLR 2014) or Stochastic Backpropagation and Approximate Inference in Deep Generative Models by Danilo Rezende (ICML 2014).
- Variational Inference, lecture note by David Blei. Section 1-6.
- Variational Inference with Normalizing Flows by Danilo Rezende (ICML 2015) and Improving Variational Inference with Inverse Autoregressive Flow by Diederik Kingma (NIPS 2016)