Robust-GAN

This repository provides an implementation of STV-GAN.

Environment

The following dependencies are needed:

• Python 3.7
• PyTorch 1.12.1

Files

This repository includes the following files,

• demo.ipynb: A Jupyter notebook demo of the code. It includes two core examples: the estimation of the mean vector and the scatter matrix.
• gan.py: This script includes a class for performing robust estimation.
• gan_torch/mu.py: This script provides a stable implementation for performing robust mean vector estimation.

Usage

1. Clone this repository to your local machine:

   git clone https://github.com/monarsan/robust-GAN
   cd robust-GAN

2. Create a new Conda environment from the environment.yaml file:

   conda env create -f environment.yaml

   This command creates a new Conda environment, which is named robust-gan.

3. Activate the new Conda environment:

   conda activate robust-gan

4. You can now start running demo.ipynb.

Citation

@article{kanamori2023robust,
  title={Robust Estimation for Kernel Exponential Families with
Smoothed Total Variation Distances},
  author={Kanamori, Takafumi and Yokoyama, Kodai and Kawashima, Takayuki},
  pages={In preparation},
}