- pytorch >=0.4.0
- torchvision ==0.2.0
- Jupyter Notebook
- A distribution of 1D MoG having three overlapping mixture components with modes at 0, 3, 6, and 110, and standard deviations of 1, 2, and 3, respectively.
[Left]: Result of Auxiliary Classifiers GAN [Right]: Result of Twin Auxiliary Classifiers GAN
- Implementing Projection cGAN.
- Implementing all the GANs with hinge loss.
- MMD evaluation for quantitative results.
- Experiments with CIFAR-100 and Overlapping MNIST
-
Twin Auxiliary Classifiers GAN. Mingming Gong*, Yanwu Xu*, Chunyuan Li, Kun Zhang, and Kayhan Batmanghelich. In Proceedings of 33rd Conference on Neural Information Processing Systems (NeurIPS 2019).(Spotlight).(https://arxiv.org/abs/1907.02690v2)