Skip to content

Python implementation of Bernoulli embeddings. Useful to create numeric representations of items in a network (e.g. trade network).

License

Notifications You must be signed in to change notification settings

yabramuvdi/bernoulli-embeddings

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

bernoulli-embeddings

This repository provides a Python implementation - relying on the JAX library- of the Bernoulli Embeddings developed by Rudolph et al. (2016). Bernoulli Embeddings are a particular class of Exponential Family Embeddings where dense low-dimensional representations of items are constructued by relying on binary co-occurence patterns. This implementation underlies current work with Vasco M. Carvalho, Stephen Hansen, and Glenn Magerman that estimates these embeddings on 14 million firm-to-firm transactions for Belgium for the year 2014.

For a complete example on how to estimate these embeddings on a particular dataset check this blog post which uses movie ratings data to construct embedded representations of movies.

About

Python implementation of Bernoulli embeddings. Useful to create numeric representations of items in a network (e.g. trade network).

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages