https://github.com/bbc/theano-bpr
An implementation of Bayesian Personalised Ranking in Theano
https://github.com/bbc/theano-bpr
Last synced: 9 months ago
JSON representation
An implementation of Bayesian Personalised Ranking in Theano
- Host: GitHub
- URL: https://github.com/bbc/theano-bpr
- Owner: bbc
- License: other
- Created: 2014-05-08T14:12:58.000Z (over 11 years ago)
- Default Branch: master
- Last Pushed: 2014-06-16T15:50:43.000Z (over 11 years ago)
- Last Synced: 2025-03-22T06:51:14.714Z (10 months ago)
- Language: Python
- Size: 211 KB
- Stars: 57
- Watchers: 26
- Forks: 23
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: COPYING
Awesome Lists containing this project
README
theano-bpr
==========
A library implementing Bayesian Personalised Ranking (BPR) for
Matrix Factorisation, as described by Rendle et al. in :
http://arxiv.org/abs/1205.2618
This model tries to predict a personalised ranking of items
from a user's viewing history. It has been shown to be
very efficient for recommendation tasks. It's also used in a variety
of other tasks, such as matrix completion, link prediction
and tag recommendation.
This library uses [Theano](http://deeplearning.net/software/theano/) and
can therefore run on a [GPU through CUDA](http://deeplearning.net/software/theano/tutorial/using_gpu.html) or on the CPU,
for which you'll need a working BLAS. We recommend using [OpenBlas](http://www.openblas.net).
Installation
------------
$ pip install theano-bpr
Usage
-----
An iPython Notebook demonstrating the use of theano-bpr over the
Movielens dataset [is available in examples/](http://nbviewer.ipython.org/github/bbcrd/theano-bpr/blob/master/examples/example.ipynb).
Testing
-------
$ nosetests
Licensing terms and authorship
------------------------------
See 'COPYING' and 'AUTHORS' files