mattscicluna / interpretable_tsne_experiment

Interpretable t-SNE Experiments

Code for reproducing the experiments found in our GLBIO oral presentation: "Towards Computing Attributions for Dimensionality Reduction Techniques"

---

Instructions

To reproduce the results from the paper, you first want to install our package from TestPyPi:

pip install -i https://test.pypi.org/simple/ interpretable-tsne

Alternatively, you can install from source using the following git repo:

https://github.com/MattScicluna/interpretable_tsne

Dont forget to install the required packages:

pip install -r requirements.txt

If you want to run the jupyter notebooks, then you will want to set up a kernel for your jupyter notebook:

python -m ipykernel install --user --name <YOUR_ENV_NAME> --display-name "Python (<YOUR_ENV_NAME>)"

---

To run the experiments

1. Download the datasets

   bash create_experiment_files.sh
   

   Note that all datasets (except for SARS-CoV-2) are available here

2. Define paths in set_environment_variables_public.sh

3. Run experiments

   bash sim_data_experiment.sh
   bash mnist_experiment.sh
   bash mnist_create_table.sh
   bash sars_cov_2_experiment.sh
   

4. Check out the Notebooks to generate the figures found in the paper!

Tutorial

We have a tutorial notebook available in notebooks/synthetic_data_tutorial.ipynb