Data Market Design through Deep Learning

This folder containts the implementation of the paper "Data Market Design through Deep Learning"
https://arxiv.org/pdf/2310.20096

Getting Started

The code is written in python and jupyter notebooks and requires the following packages:

• Numpy
• Scipy
• Matplotlib
• Pytorch

Running the experiments

Single-Buyer Settings

The notebooks in the RochetNet folder has all jupyter notebooks needed to recover the results in the main paper and appendix. The hyperparamters are set already in the cell that defines the config class. To recover the results that show how differential informativeness varies for the enlarged buyer type, go to RochetNet/enlarged_types/Enlarged_types_mix.ipynb and change the values corresponding to cfg.v_L to vary a and re-run the notebook.

Multi-Buyer Settings

BIC Settings

The notebooks in the BIC folder has all jupyter notebooks needed to recover the results in the main paper and appendix. The hyperparamters are set already in the cell that defines the config class. To recover the results computing how revenue varies $\alpha$, go to BIC/uncertain_theta/UNF_NE.ipynb and BIC/uncertain_v/U_theta_0.50_alpha_0.50.ipynb and set cfg.alpha to the appropriate value and re-run the experiments

Ex post IC settings

The notebooks in the IC folder has all jupyter notebooks needed to recover the results in the main paper and appendix. The hyperparamters are set already in the cell that defines the config class.

Reference

Please cite our work if you find our code/paper is useful to your work.

@inproceedings{ravindranath2023data,
  title={Data Market Design through Deep Learning},
  author={Sai Srivatsa Ravindranath and Yanchen Jiang and David C. Parkes},
  booktitle={Thirty-seventh Conference on Neural Information Processing Systems},
  year={2023},
  url={https://openreview.net/forum?id=sgCrNMOuXp}
}