Skip to content
This repository has been archived by the owner on May 26, 2021. It is now read-only.

ubisoft/ubisoft-laforge-disentanglement-metrics

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Supervised Disentanglement Metrics

This repository contains code for the paper:

J. Zaidi, J. Boilard, G. Gagnon and M.A. Carbonneau, "Measuring Disentanglement: A Review of Metrics", arXiv:2012.09276, 2020.

Please cite this paper if you use the code in this repository as part of a published research project.

Setup

The code was run using python 3.8:

  1. create a python virtual environment
  2. clone this repo: git clone https://github.com/Ubisoft-LaForge/ubisoft-laforge-DisentanglementMetrics.git
  3. navigate to the repository: cd disentanglement_metrics
  4. install python requirements: pip install -r requirements.txt

Disentanglement Metrics

All 12 disentanglement metrics studied in the paper are implemented in src/metrics/ folder.

Each metric implementation takes as input:

  1. a set of factors of shape (nb_examples, nb_factors)
  2. a set of codes of shape (nb_examples, nb_codes)
  3. additional hyper-parameters specific to the metric

The default hyper-parameters used for each metric in the paper are in script src/experiments/config.py

Reproducing The Results

We provide the code that was used to compute results from experiments of Section 5.2 to 5.6. All the scripts are in src/experiments/ folder. Each script can be run in 2 modes:

  • run: get scores for each metric and save them
  • plot: plot scores for each metric family and save plots

Metric scores and plots will be automatically saved into results/ folder, at the root of the repository.

For example, to reproduce noise experiment of section 5.2:

  1. navigate to experiments folder: cd src/experiments
  2. compute metric scores: python section5.2_noise.py run
  3. plot scores: python section5.2_noise.py plot

NOTES:

  • By default, all metrics scores are computed. It is possible to target specific metrics as follows: python section5.2_noise.py run --metrics "metric_1" ... "metric_N"
  • For example, to only compute scores for predictor-based metrics: python section5.2_noise.py run --metrics "Explicitness" "SAP" "DCI Lasso" "DCI RF"

Feedback

Please send any feedback to marc-andre.carbonneau2@ubisoft.com and julian.zaidi@ubisoft.com.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages