Valuation Meets RTBF

Code for implementation of "Equitable Data Valuation Meets the Right to Be Forgotten in Model Markets".

Warning

The code is still untidied and may exist bugs.

Prerequisites

• Python, NumPy, Scikit-learn, PyTorch

Experiments in the Paper

They can be found in folder paper_exps.

Basic Usage

To divide value fairly between data owners with a given sharded structure, the learning algorithm and a measure of performance for the trained model (test accuracy, etc.).

License

This project is licensed under the MIT License - see the LICENSE file for details.