PS-MI

Introduction

This is a repository for the paper "PS-MI: Accurate, Efficient, and Private Data Valuation in Vertical Federated Learning".

Requirements

grpcio==1.34.1
grpcio-tools==1.34.1
tenseal==0.4.0
numpy==1.19.5
pandas==1.1.5
scikit-learn==0.24.1
torch==1.8.1

Parameters

The parameters are defined in the conf/args.py file.

How to Run

1. Data Preparation:

• Download the dataset from: UCI, Kaggle website
• Put the dataset in the data folder.
• The tools for data preprocessing are in the data_loader folder.

2. Run the scripts:

• script/mi)shapley/PS-Mi.sh: Run the PS-Mi algorithm.