NeuralFDR

Software accompanying "NeuralFDR: Learning Discovery Thresholds from Hypothesis Features", NIPS 2017

Dependencies

You will have to install PyTorch to run the code, follow the instructions from http://pytorch.org

Download the data

Download the data used in the paper from this dropbox folder.

Train a NeuralFDR model

python train.py --data data/data_airway.csv --dim 1 --out airway

The report will be available in airway folder

Citation

If you use this code, please cite

@inproceedings{xia2017neuralfdr,
  title={NeuralFDR: Learning Discovery Thresholds from Hypothesis Features},
  author={Xia, Fei and Zhang, Martin J and Zou, James Y and Tse, David},
  booktitle={Advances in Neural Information Processing Systems},
  pages={1540--1549},
  year={2017}
}