google-research

Introduction

Tensorflow implementation of Teaching with Fairness project

Requirements

1. Python 3.6
2. TensorFlow 1.14.0
3. scikit-image

Usage

Part 1: Download the dataset from google drive

Google Drive

Part 2: Train and test with the following code in different folder.

# train and generate fake images with GAN
cd gan
python train.py
python generate.py

# train the teacher and student model with RL
cd rl
python train.py
python restore.py

# compare with the baseline 
cd baseline
python all_real.py #train with all real data
python all_fake.py #train with all fake data
python random.py #train with random combination of real and fake data
python balance.py #train with balanced combination of real and fake data