GAN implementations
Switch branches/tags
Nothing to show
Clone or download
Permalink
Failed to load latest commit information.
01 vanilla gan Add DCGAN Feb 6, 2018
02 conditional gan Add anogan dev Jan 17, 2018
03 dcgan haha Feb 9, 2018
04 rnn gan Add RNN GAN Feb 10, 2018
05 improved gan Modify Feature Matching Feb 14, 2018
07 least squares gan Add DCGAN Feb 6, 2018
08 anogan Delete Anomalous Images Feb 26, 2018
images Complete AnoGAN Feb 26, 2018
.gitignore Delete Anomalous Images Feb 26, 2018
03-gan-list.ipynb dev Jan 12, 2018
README.md Complete AnoGAN Feb 26, 2018

README.md

gan-yhat

GAN implementations with PyTorch

The followings are a list of GAN implemented here.

  1. Vanilla GAN
    • Vanilla GAN for MNIST
    • Vanilla GAN for Fashion MNIST
  2. Conditional GAN for MNIST
  3. DCGAN for Fashion MNIST
  4. Improved GAN
    • Feature Maching
  5. AnoGAN

Vanilla GAN

MNIST

It shows a little bit of mode collapse; a commonly encountered failure case for GANs where the generator produces samples with extremely low variety. In this case, the generator produces 1 with extremely high probability.

MNIST

Conditional GAN

Conditional GAN reduces mode collapse issue by giving the model additional information.

Conditional GAN

DCGAN

DCGAN makes use of convolutions and transposed convolutions.

Conditional GAN

Conditional GAN

Conditional GAN

RNN GAN

Sin

It uses feature mapping and LSTM.

Sin

Improved GAN

Feature Matching

Conditional GAN

Conditional GAN

AnoGAN

Anomaly Detection with GAN

images generated by GAN Generator

The following images are generated by the GAN Generator. It is the same model as the DCGAN model.

Normal1

Anomaly Detection for Normal Data

The following images are normal images.

It is good if red dots are less shown.

Normal1

Normal1

Normal1

Normal1

Normal1

Normal1

Normal1

Anomaly Detection for Abnormal Data

Abnormal1

Abnormal1

Abnormal1

Abnormal1

Abnormal1

Abnormal1

Abnormal1

Abnormal1