pyro-ppl/pyro

Stick-breaking variational autoencoder example

Open

#1,339 opened on Aug 26, 2018

View on GitHub
 (1 comment) (0 reactions) (0 assignees)Python (8,211 stars) (981 forks)batch import
Examplesdiscussionhelp wanted

Description

I'd like to add a new example for Nalisnick and Smyth's Stick-Breaking Variational Autoencoder.

Since @fritzo and @martinjankowiak have implemented pathwise gradients for the beta distribution, what would be a good first step in this direction?

Contributor guide

Tech stack
python
Domain
machine learning
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-2 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
Pyro basicsVAE understandingbeta distributionresearch paper knowledge
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
30
Research direction
Start by reviewing existing VAE examples in the Pyro repository (e.g., examples/vae/). The stick breaking VAE uses a prior based on a stick breaking process with beta distributions; leverage the pathwise gradients for beta distributions already implemented by @fritzo and @martinjankowiak. Implement the encoder and decoder following the paper's architecture, and consider using Pyro's SVI with ELBO. Test on a simple dataset like MNIST.
Stick-breaking variational autoencoder example · pyro-ppl/pyro#1339 | Good First Issue