MMGAN: Generative Adversarial Networks for Multi-Modal Distributions
This addresses the issue of handling heterogeneous data distributions in generative models, particularly for applications requiring clustering, but it is incremental as it builds on existing GAN frameworks.
The paper tackles the problem of unstable training and mode collapse in GANs for multi-modal data by proposing MMGAN, which models the latent space as a Gaussian mixture model and includes a clustering network, resulting in improved clustering performance over state-of-the-art models in benchmark experiments.
Over the past years, Generative Adversarial Networks (GANs) have shown a remarkable generation performance especially in image synthesis. Unfortunately, they are also known for having an unstable training process and might loose parts of the data distribution for heterogeneous input data. In this paper, we propose a novel GAN extension for multi-modal distribution learning (MMGAN). In our approach, we model the latent space as a Gaussian mixture model with a number of clusters referring to the number of disconnected data manifolds in the observation space, and include a clustering network, which relates each data manifold to one Gaussian cluster. Thus, the training gets more stable. Moreover, MMGAN allows for clustering real data according to the learned data manifold in the latent space. By a series of benchmark experiments, we illustrate that MMGAN outperforms competitive state-of-the-art models in terms of clustering performance.