Multi-Generator Generative Adversarial Nets

By Quan Hoang, Tu Dinh Nguyen, Trung Le, Dinh Q. Phung. 2017

69 citations recorded by the source corpus.

We propose a new approach to train the Generative Adversarial Nets (GANs) with a mixture of generators to overcome the mode collapsing problem. The main intuition is to employ multiple generators, instead of using a single one as in the original GAN. The idea is simple, yet proven to be extremely effective at covering diverse data modes, easily overcoming the mode collapse and delivering state-of-the-art results. A minimax formulation is able to establish among a classifier, a discriminator, and a set of generators in a similar spirit with GAN. Generators create samples that are intended to come from the same distribution as the training data, whilst the discriminator determines whether samples are true data or generated by generators, and the classifier specifies which generator a sample comes from. The distinguishing feature is that internal samples are created from multiple generators, and then one of them will be randomly selected as final output similar to the mechanism of a probabilistic mixture model. We term our method Mixture GAN (MGAN). We develop theoretical analysis to prove that, at the equilibrium, the Jensen-Shannon divergence (JSD) between the mixture of generators' distributions and the empirical data distribution is minimal, whilst the JSD among generators' distributions is maximal, hence effectively avoiding the mode collapse. By utilizing parameter sharing, our proposed model adds minimal computational cost to the standard GAN, and thus can also efficiently scale to large-scale datasets. We conduct extensive experiments on synthetic 2D data and natural image databases (CIFAR-10, STL-10 and ImageNet) to demonstrate the superior performance of our MGAN in achieving state-of-the-art Inception scores over latest baselines, generating diverse and appealing recognizable objects at different resolutions, and specializing in capturing different types of objects by generators.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Wireless Network Intelligence at the Edge (2018, background): “In the opposite way, multiple generators can be distributed over the devices that share a single discriminator [205].”
Online Adaptative Curriculum Learning for GANs (2018, background): “2018), multiple discriminators (Neyshabur, Bhojanapalli, and Chakrabarti 2017; Juefei-Xu, Boddeti, and Savvides 2017), or multiple generators (Tolstikhin et al. 2017; Hoang et al. 2017; Kwak and Zhang 2016).”
Dual Discriminator Generative Adversarial Nets (2017, background): “The third approach is to train many generators that discover different modes of the data [31].”
Generative Adversarial Networks (GANs) (2020, methodology): “Training of Multiple Gs (S12(ii)) cGAN [50], AdaGAN [51], MAD-GAN [67], MGAN [68], MPM GAN [69], FictitiousGAN [70], MIX+GAN [71]”
A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications (2020, background): “3.3.3.9 Multi-generator learning: Multi-generator GAN (MGAN) [45] is proposed to train the GANs with a mixture of generators to avoid the mode collapsing problem.”
Dropout-GAN: Learning from a Dynamic Ensemble of Discriminators (2018, background): “don, UK (a) Toy dataset. (b) Wasserstein distance. (c) Symmetric KL divergence. Figure 8: Comparison of Dropout-GAN (8 discriminators) with original GANs [15], Unrolled GAN [28], D2GAN [30], and MGAN [19] (a). Real data is presented in red while generated data is presented in blue. Our method manage”
A Comprehensive Survey of AI-Generated Content (AIGC): A History of Generative AI from GAN to ChatGPT (2023, background): “MGAN [82] and MAD-GAN [83] address the mode collapse problem by incorporating multiple generators and one discriminator.”