Python Tensor Decomposition Algorithms

This repository implements efficient numerical algorithm for Alternating Least Squares (ALS) in CP and Tucker decompositions, Pairwise perturbation algorithm for CP-ALS as well as fast Nonlinear Least Squares (NLS) for CP decomposition.

It also implements Alternating Mahalanobis Distance Minimization and hybrid updates for CP decomposition

This repository implements everything in Python, and is compatible with both Numpy backend, which allows fast seaquential running, and Cyclops Tensor Framework backend, which allows fast distributed parallism.

Prerequisite

Run

pip install -r requirements.txt

to install necessary packages.

Tests cases

Run all tests with

# sudo pip install nose
nosetests -v tests/*.py

Running CP/Tucker decomposition with ALS

Run

python run_als.py -h

to see the existing input arguments and their functions.

Run

python tests.py

to test some simple CP runs.

Running CP decomposition with NLS

Run

python run_nls.py -h

to see existing input arguments and their functions with default values

CPD with AMDM and hybrid algorithms

python mahalanobis.py -h

to see existing input arguments and their functions with default values

ALS/GN performance comparision

The comparison in contraction ALS and GN CG iteration is done using Contraction.py file. Run

python Contraction.py -h

to see existing input arguments

Convergence probability

The file Convprob.py can be used to compare the convergence probability of different methods. Run

python Convprob.py -h

to see input arguments with their functions and default values

Visualization with Visdom

For now visdom can fetch all the csv files following the particular format and plot them.

Go to the Visdom folder then execute the following commands:

visdom -port XXXXX

python visdom_pull_server.py -port XXXXX