https://github.com/andymiller/vboost

code supplement for variational boosting (https://arxiv.org/abs/1611.06585)
https://github.com/andymiller/vboost

Last synced: about 2 months ago
JSON representation

code supplement for variational boosting (https://arxiv.org/abs/1611.06585)

• Host: GitHub
• URL: https://github.com/andymiller/vboost
• Owner: andymiller
• License: mit
• Created: 2017-06-06T17:45:54.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2017-07-24T17:03:53.000Z (about 7 years ago)
• Last Synced: 2024-07-02T16:42:12.548Z (3 months ago)
• Language: Python
• Homepage:
• Size: 11.3 MB
• Stars: 11
• Watchers: 6
• Forks: 5
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-gradient-boosting-papers - [Code

README

        
# vboost

code for [Variational Boosting: Iteratively Refining Posterior Approximations](https://arxiv.org/abs/1611.06585)

### Abstract

> We propose a black-box variational inference method to approximate

> intractable distributions with an increasingly rich approximating class.

> Our method, termed variational boosting, iteratively refines an existing

> variational approximation by solving a sequence of optimization problems,

> allowing the practitioner to trade computation time for accuracy.

> We show how to expand the variational approximating class by incorporating

> additional covariance structure and by introducing new components to form a

> mixture. We apply variational boosting to synthetic and real statistical

> models, and show that resulting posterior inferences compare favorably to

> existing posterior approximation algorithms in both accuracy and efficiency.

Authors:

[Andrew Miller](http://andymiller.github.io/),

[Nick Foti](http://nfoti.github.io/), and

[Ryan Adams](http://people.seas.harvard.edu/~rpa/).

### Requires

* [`autograd`](https://github.com/HIPS/autograd) + its requirements (`numpy`, etc).

  Our code is compatible with [this `autograd` commit](https://github.com/HIPS/autograd/tree/42a57226442417785efe3bd5ba543b958680b765) or later.

  You can install the master version with

  `pip install git+git://github.com/HIPS/autograd.git@master`.

* [`pyprind `](https://github.com/rasbt/pyprind)

* [`sampyl`](https://github.com/mcleonard/sampyl) for MCMC experiments