Abstract:Original data usually lies on a set of disconnected manifolds rather than a smooth connected manifold. This causes the problem of mode collapse in the training of vanilla Generative Adversarial Network (GAN). There are many existing GAN variants that attempt to address this problem, but they result in limitations. The existing variants either produce simulated instances with low quality or generate identical simulated instances. In this study, we propose a new approach to training GAN utilizing multiple generators, a classifier and a discriminator to address mode collapse. The classifier outputs the statistical probabilities of generated data belonging to a specific category. These probabilities implicitly reflect which manifolds are captured by generators, and the correlation between generators is quantified by mutual information. Our idea views the mutual information values as a constraint to guide generators in learning different manifolds. Specifically, we traverse the generators, calculating the mutual information between each generator and the others. The calculated values are integrated into the generator loss to form a new generator loss and to update the corresponding generator's parameters, using back-propagation. We minimize the mutual information to reduce the correlation between generators while also minimizing the generator loss. This ensures generators capture different manifolds while updating their parameters. A new minimax formula is established to train our approach in a similar spirit to vanilla GAN. We term our approach Mutual Information Multi-generator GAN (MIM-GAN). We conduct extensive experiments utilizing the MNIST, CIFAR10 and CelebA datasets to demonstrate the significant performance improvement of MIM-GAN in both achieving the highest Inception Scores and producing diverse generated data at different resolutions.

Improving mode exploring capability of generative adversarial nets by self-organizing map

Dual Distribution Matching GAN

Soft Generative Adversarial Network: Combating Mode Collapse in Generative Adversarial Network Training Via Dynamic Borderline Softening Mechanism

Towards the Gradient Vanishing, Divergence Mismatching and Mode Collapse of Generative Adversarial Nets

On Mode Collapse in Generative Adversarial Networks

Combating Mode Collapse in GANs via Manifold Entropy Estimation

Combating Mode Collapse via Offline Manifold Entropy Estimation

Multi-generator GAN learning disconnected manifolds with mutual information

VirtualGAN: Reducing Mode Collapse in Generative Adversarial Networks Using Virtual Mapping

Overcoming Mode Collapse with Adaptive Multi Adversarial Training

Improving GAN with Neighbors Embedding and Gradient Matching

Optimizing Generative Adversarial Networks in Latent Space

No Modes Left Behind: Capturing the Data Distribution Effectively Using GANs

Study of Prevention of Mode Collapse in Generative Adversarial Network (GAN)

Latent Dirichlet Allocation Based Generative Adversarial Networks.

Mode Regularized Generative Adversarial Networks

Alleviating Mode Collapse in GAN via Diversity Penalty Module

BourGAN: Generative Networks with Metric Embeddings

Hierarchical Modes Exploring in Generative Adversarial Networks