Time: Tuesday, 2:00 pm - 3:00 pm (once every other week or by demand)
Location: Fishbowl
Bayesian modeling of massive data is computationally challenging and some algorithms such as ABC and MCMC are notoriously expensive. Variational Autoencoders and Variational Bayes attempt to speed computation and are gaining popularity.
Schedule (Open an issue if you want to suggest a paper to read):
| Date | Presentation | Presenter | Summary |
|---|---|---|---|
| October 8 | Hierarchical Decompositional Mixtures of Variational Autoencoders | Julia | We discussed background on Autoencoders and Variational Autoencoders and very briefly the SPN extra layer (hierarchy) in the VAE proposed in the paper. |
| October 22 | Information Constraints on Auto-Encoding Variational Bayes | Jaehee | |
| November 5 | Hierarchical Representations with Poincaré Variational Auto-Encoders | Airam | |
| November 19 | Tutorial on Variational Autoencoders by Scott Linderman | ||
| December 3 | Bayesian cluster analysis | Lorenzo | |
| Later | Model criticism in latent space | ||
| Later | Augmented Neural ODEs | ||
| Later | Informed Proposals for Local MCMC in Discrete Spaces |
Put yourself on the calendar like this:
| May 16 | [paper](https://...) | Julia |