[TensorFlow] Wasserstein GAN (WGAN)

TensorFlow implementation of Wasserstein GAN (WGAN) with MNIST dataset.

Training algorithm

The algorithm for training WGAN [1].

WGAN architecture

The architecture of WGAN [1].

Graph in TensorBoard

Graph of WGAN.

Results

Training Procedure

Loss graph in the training procedure.
Each graph shows loss of the discriminator and loss of the generator respectively.

Test Procedure

z:2	z:2 (latent space walking)

z:64	z:128

Environment

• Python 3.7.4
• Tensorflow 1.14.0
• Numpy 1.17.1
• Matplotlib 3.1.1
• Scikit Learn (sklearn) 0.21.3

Reference

[1] Martin Arjovsky et al. (2017). Wasserstein GAN. arXiv preprint arXiv:1701.07875.