/ebgan

A tensorflow implementation of Junbo et al's Energy-based generative adversarial network ( EBGAN ) paper.

Primary LanguagePythonMIT LicenseMIT

EBGAN

A tensorflow implementation of Junbo et al's Energy-based generative adversarial network ( EBGAN ) paper. ( See : https://arxiv.org/pdf/1609.03126v2.pdf ) My implementation is somewhat different from original papers, for example I've used convolution layers in both generator and discriminator instead of fully connected layers. I think this isn't important and will not make a big difference in the final result.

Version

Current Version : 0.0.0.2

Dependencies ( VERSION MUST BE MATCHED EXACTLY! )

  1. tensorflow == 1.0.0
  2. sugartensor == 1.0.0.2

Training the network

Execute


python mnist_ebgan_train.py

to train the network. You can see the result ckpt files and log files in the 'asset/train' directory. Launch tensorboard --logdir asset/train/log to monitor training process.

Generating image

Execute


python mnist_ebgan_generate.py

to generate sample image. The 'sample.png' file will be generated in the 'asset/train' directory.

Generated image sample

This image was generated by EBGAN network.

Other resources

  1. Original GAN tensorflow implementation
  2. InfoGAN tensorflow implementation
  3. Supervised InfoGAN tensorflow implementation

Authors

Namju Kim (buriburisuri@gmail.com) at Jamonglabs Co., Ltd.