Mixed Membership Generative Adversarial Networks

Yasin Yazici, Bruno Lecouat, Kim-Hui Yap, Stefan Winkler, Georgios Piliouras, Vijay Chandrasekhar, Chuan-Sheng Foo

Abstract

GANs are designed to learn a single distribution, though multiple distributions can be modeled by treating them separately. However, this naive implementation does not consider overlapping distributions. We propose Mixed Membership Generative Adversarial Networks (MMGAN) analogous to mixed-membership models that model multiple distributions and discover their commonalities and particularities. Each data distribution is modeled as a mixture over a common set of generator distributions, and mixture weights are automatically learned from the data. Mixture weights can give insight into common and unique features of each data distribution. We evaluate our proposed MMGAN and show its effectiveness on MNIST and Fashion-MNIST with various settings.
BibTeX
@inproceedings{yazici2022mixed,
  author = {Yasin Yazici and Bruno Lecouat and Kim-Hui Yap and Stefan Winkler and Georgios Piliouras and Vijay Chandrasekhar and Chuan-Sheng Foo},
  title = {{Mixed Membership Generative Adversarial Networks}},
  booktitle = {ICIP},
  year = {2022},
  doi = {10.1109/icip46576.2022.9897640},
}

Loading related content...

Figure