Repository for the draft `A FAST MULTIPLICATIVE UPDATES ALGORITHM FOR NON-NEGATIVE MATRIX FACTORIZATION'

This is a temporary repository to share the codes for the submitted draft on fastMU. The code will be improved upon publication.

Organisation

• The following scripts contain the algorithms
  • NLS_Frobenius
  • NLS_KL
  • NMF_Frobenius
  • NMF_KL
• The experiments are in the scripts
  • "synthetic_comparisons_Frobenius_nls.py",
  • "synthetic_comparisons_Frobenius.py",
  • "synthetic_comparisons_KL_nls_sparse.py",
  • "synthetic_comparisons_KL_sparse.py",
  • "synthetic_Frobenius_delta_choice.py",
  • "synthetic_KL_delta_choice.py",
  • "test_audio_nls.py",
  • "test_audio_nmf.py",
  • "test_hyperspectral_nls.py",
  • "test_hyperspectral.py"
• Data is stored in data_and_scripts/
• All results including figures are stored in Results/

Installation

The following libraries are required to run the algorithms:

• numpy
• scipy
• matplotlib

Additionally, to run the experiments, install the following packages:

• plotly
• pandas
• soundfile (audio experiment only)

Two custom packages are also required:

• nnfac (local version): install by

  pip install .
  

  in the nnfac_perso directory
• shootout: install from this github repository, should work with commit 511974955b4e458608ef6ebd3c0a5db2e12cdcf5

Running experiments

Simply run the runscript.py script. It take as input parameter the number of random samples for the comparisons:

python runscript.py 0

uses local stored result to produce figures

python runscript.py xx

with xx some natural number runs all the experiments with xx samples, stores the results and produces figures.