DeePINK - Bigger Net Evaluation

An add-on for DeePINK paper implementation, using a bigger, more complex CNN and evaluating DeePINK`s strengh with it.

Offry Hayon • Tomer Keren

Running the program

1. Download repo either by pressing "Download ZIP" under the green button or use clone command

git clone https://github.com/tomerkeren42/DeePINK-BiggerNet.git

2. Install requriements:

pip install -r requirements.txt

3. Run python quickstart.py OR use the attached script for running the program, start_evaluate_deePINK_extension.py

python quickstart.py

OR

python start_evaluate_deePINK_extension.py

4.Enjoy!

References:

Original DeePINK paper

@inproceedings{Lu2018DeepPINKRF,
 title={DeepPINK: reproducible feature selection in deep neural networks},
 author={Yang Young Lu and Yingying Fan and Jinchi Lv and William Stafford Noble},
 booktitle={NeurIPS},
 year={2018}
}

Original KnockPy implementation

@article{AS-LJ:2020,
 title={Powerful Knockoffs via Minimizing Reconstructability},
 author={Spector, Asher and Janson, Lucas},
 journal={Annals of Statistics},
 year={2021+},
 note={To Appear}
}