/LSC

Code for the arXiv preprint "Disentangling images with Lie group transformations and sparse coding" (2020).

Primary LanguageJupyter NotebookMIT LicenseMIT

Lie Group Sparse Coding

Code for the arXiv preprint "Disentangling images with Lie group transformations and sparse coding" (2020).

@article{chau2020disentangling,
  title={Disentangling images with Lie group transformations and sparse coding},
  author={Chau, Ho Yin and Qiu, Frank and Chen, Yubei and Olshausen, Bruno},
  journal={arXiv preprint arXiv:2012.12071},
  year={2020}
}

To try the code, open lsc.ipynb and run the cells.

Requires pytorch, numpy, scipy, and matplotlib. By default, uses GPU.

We thank Taco Cohen for providing the code for the paper: "Learning the Irreducible Representations of Commutative Lie Groups" and the code helped us a lot during this project. If you find this code useful, please also consider to cite the following work:

@inproceedings{cohen2014learning,
  title={Learning the irreducible representations of commutative lie groups},
  author={Cohen, Taco and Welling, Max},
  booktitle={International Conference on Machine Learning},
  pages={1755--1763},
  year={2014},
  organization={PMLR}
}