/valuationgame

Primary LanguageJupyter NotebookMIT LicenseMIT

Energy-Based Learning for Cooperative Games, with Applications to Valuation Problems in Machine Learning

This repository contains code for the paper: "Energy-Based Learning for Cooperative Games, with Applications to Valuation Problems in Machine Learning".

Run the experiments

Synthetic experiments with FLID games

Run the file synthetic/main_syn.ipynb directly.

Data valuation

Run the following bash file:

data_valuation/run_data_val.sh

Feature valuation

Run the ipython notebook feature_valuation/feature_val_example.ipynb.

Main depencencies (has been tested on Linux)

  • python>=3.5
  • shap==0.39.0
  • tensorflow>2.0
  • scikit-learn==0.24
  • xgboost
  • numba==0.53.1
  • tqdm
  • matplotlib

Reference

😄If you find this repo is useful, please consider to cite our paper:

@inproceedings{
bian2022energybased,
title={Energy-Based Learning for Cooperative Games, with Applications to Valuation Problems in Machine Learning},
author={Yatao Bian and Yu Rong and Tingyang Xu and Jiaxiang Wu and Andreas Krause and Junzhou Huang},
booktitle={International Conference on Learning Representations},
year={2022},
url={https://openreview.net/forum?id=xLfAgCroImw}
}

Disclaimer

This is not an officially supported Tencent product.