https://github.com/caselles/neurips19-sbdrl

Code for NeurIPS 2019 paper: "Symmetry-Based Disentangled Representation Learning requires Interaction with Environments" by H. Caselles-Dupré, M. Garcia-Ortiz and D. Filliat.
https://github.com/caselles/neurips19-sbdrl

Last synced: 11 months ago
JSON representation

Code for NeurIPS 2019 paper: "Symmetry-Based Disentangled Representation Learning requires Interaction with Environments" by H. Caselles-Dupré, M. Garcia-Ortiz and D. Filliat.

• Host: GitHub
• URL: https://github.com/caselles/neurips19-sbdrl
• Owner: Caselles
• License: mit
• Created: 2019-09-11T12:40:41.000Z (almost 7 years ago)
• Default Branch: master
• Last Pushed: 2019-12-09T01:14:47.000Z (over 6 years ago)
• Last Synced: 2025-06-03T01:51:28.841Z (about 1 year ago)
• Language: Python
• Homepage:
• Size: 6.11 MB
• Stars: 35
• Watchers: 4
• Forks: 11
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Code for NeurIPS 2019 paper.

   "Symmetry-Based Disentangled Representation Learning requires Interaction with Environments"  


This repository contains the official code for the NeurIPS 2019 paper "Symmetry-Based Disentangled Representation Learning requires Interaction with Environments" by Hugo Caselles-Dupré, Michael Garcia-Ortiz and David Filliat.

> **Symmetry-Based Disentangled Representation Learning requires Interaction with Environments**


> Hugo Caselles-Dupré ¹ ², Michael Garcia-Ortiz ² and David Filliat ¹


>¹ Flowers Laboratory (ENSTA Paris   & INRIA )  


>² Softbank Robotics Europe  


> https://arxiv.org/abs/1904.00243

>

> **Abstract:** *Finding a generally accepted formal definition of a disentangled representation in the context of an agent behaving in an environment is an important challenge towards the construction of data-efficient autonomous agents. Higgins et al. recently proposed Symmetry-Based Disentangled Representation Learning, a definition based on a characterization of symmetries in the environment using group theory. We build on their work and make observations, theoretical and empirical, that lead us to argue that Symmetry-Based Disentangled Representation Learning cannot only be based on fixed data samples. Agents should interact with the environment to discover its symmetries.*



  


  Open Colab Notebook to reproduce paper's experiments.



------------------------

## Resources

All material related to our paper is available via the following links:

| Link                      | Description

| :--------------           | :----------

| [Link ArXiv](https://arxiv.org/abs/1904.00243) | Paper PDF.

| [Link Project Page](https://sites.google.com/view/sbdrl-neurips19/) | Project page.

| [Link Video](https://youtu.be/jyN3ZQNH1GI) | Summary video.

| [Link Colab](https://colab.research.google.com/drive/1KVlSV24c687N_4TLJWwGTkjt3sh9ufWW) | Colab to reproduce experiments.

| [Link Github](https://github.com/Caselles/Symmetry_based_Disentanglement)  | Source code.

| [Link Slides](https://github.com/Caselles/NeurIPS19-SBDRL/blob/master/static/%5BNeurIPS19%5D%20SBDRL%20requires%20Interactions%20with%20Environments.pdf) | Presentation slides.

| [Link Poster](https://github.com/Caselles/NeurIPS19-SBDRL/blob/master/static/_NeurIPS19_Poster__SBDRL_requires_Interactions_w_Env.pdf) | Poster.