https://github.com/zsteve/wtf
wasserstein tensor factorisation
https://github.com/zsteve/wtf
Last synced: 8 months ago
JSON representation
wasserstein tensor factorisation
- Host: GitHub
- URL: https://github.com/zsteve/wtf
- Owner: zsteve
- License: mit
- Created: 2021-04-05T16:01:39.000Z (about 5 years ago)
- Default Branch: main
- Last Pushed: 2021-08-18T20:07:31.000Z (almost 5 years ago)
- Last Synced: 2025-03-29T09:24:09.850Z (about 1 year ago)
- Language: Roff
- Size: 108 MB
- Stars: 6
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# wtf: Wasserstein Tensor Factorisation
A unified framework for non-negative matrix and tensor factorisations with a smoothed Wasserstein loss
This repository contains a basic implementation of the method described in the article "A unified framework for non-negative matrix and tensor factorisations with a smoothed Wasserstein loss".
## Requirements
- PyTorch
- [Tensorly](http://tensorly.org)
- [PythonOT](https://pythonot.github.io/)
- CUDA-compatible GPU (e.g. use [Colab](http://colab.research.google.com/)) for efficient autodiff
## Instructions
- Clone this repo: `git clone https://github.com/zsteve/wtf.git`
- Import the `wtf` module using
```python
import sys
sys.path.insert(0, "/content/wtf/src")
import wtf
```
- ???
- Profit
## Example
### From the paper
A notebook for Figures 3, 4, 5 is located in the `examples/` directory.
[](https://colab.research.google.com/github/zsteve/wtf/blob/main/examples/example.ipynb)
### Amino acids fluorescence dataset
A notebook for the [amino acids fluorescence dataset](http://www.models.life.ku.dk/Amino_Acid_fluo) of Andersson and Bro is available as `examples/amino.ipynb`.
Thanks to Shoaib Bin Masud (Tufts) and Anna Konstorum (Yale) for bringing this dataset to my attention.
[](https://colab.research.google.com/github/zsteve/wtf/blob/main/examples/amino.ipynb)
## Citing
- If you find this work relevant to your research project, please cite the [preprint](https://arxiv.org/abs/2104.01708)
```
Zhang, S. A unified framework for non-negative matrix and tensor factorisations with a smoothed Wasserstein loss, arXiv preprint arXiv:2104.01708, 2021
```