https://github.com/stitchfix/ntflib

Sparse Beta-Divergence Tensor Factorization Library
https://github.com/stitchfix/ntflib

Last synced: about 1 year ago
JSON representation

Sparse Beta-Divergence Tensor Factorization Library

• Host: GitHub
• URL: https://github.com/stitchfix/ntflib
• Owner: stitchfix
• License: mit
• Created: 2015-04-12T01:16:55.000Z (over 11 years ago)
• Default Branch: master
• Last Pushed: 2015-10-28T22:34:26.000Z (over 10 years ago)
• Last Synced: 2025-03-30T13:03:47.769Z (over 1 year ago)
• Language: Python
• Size: 11.5 MB
• Stars: 48
• Watchers: 98
• Forks: 9
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE
  • Code of conduct: CODE_OF_CONDUCT.md

Awesome Lists containing this project

README

          
# NTFLib

Sparse Beta-Divergence Tensor Factorization Library

![Build Status](https://api.travis-ci.org/stitchfix/NTFLib.svg)

![NTFLib](decomposition.png)

Based off of this [beta-NTF project](https://code.google.com/p/beta-ntf/) this library is 

specially-built to handle tensors where the sparsity implies missing data. Most

libraries are formulated around the handling of images which are dense or nearly

dense. However, we're interested in factorizing highly sparse (>99.9%) tensors

and have written NTFLib for this use case.

Currently, we only support rank-3 tensors, but extending to this to higher rank

tensors is fairly straightforward.

The algorithm implemented by beta_ntf features standard multiplicative updates

minimizing β-divergence, which were recently shown to guarantee a decrease of

the cost-function at each step. For recent references on the topic, check for

example:

- A. Cichocki, R. Zdunek, A. H. Phan, and S. Amari, Nonnegative matrix and tensor

factorizations : Applications to exploratory multi-way data analysis and blind

source separation, Wiley Publishing, September 2009.

- C. Févotte and J. Idier, Algorithms for nonnegative matrix factorization with

the beta-divergence, Neural Computation 23 (2011), no. 9, 2421–2456.

- Virtanen, Tuomas, Monaural Sound Source Separation by Perceptually Weighted

Non-Negative Matrix Factorization, Technical report, Tampere University of

Technology, Institute of Signal Processing, 2007.