Lessons Learned from the Training of GANs on Artificial Datasets
This addresses the problem of GAN training instability and mode collapse for researchers, offering insights that are incremental but with strong experimental validation.
The paper tackles the difficulty of analyzing GANs due to dataset issues by training them on artificial datasets with infinite samples and simple distributions, finding that generators fail to learn optimal parameters and that mixtures of GANs outperform depth or width increases, with a mixture on CIFAR-10 significantly beating state-of-the-art in IS and FID metrics.
Generative Adversarial Networks (GANs) have made great progress in synthesizing realistic images in recent years. However, they are often trained on image datasets with either too few samples or too many classes belonging to different data distributions. Consequently, GANs are prone to underfitting or overfitting, making the analysis of them difficult and constrained. Therefore, in order to conduct a thorough study on GANs while obviating unnecessary interferences introduced by the datasets, we train them on artificial datasets where there are infinitely many samples and the real data distributions are simple, high-dimensional and have structured manifolds. Moreover, the generators are designed such that optimal sets of parameters exist. Empirically, we find that under various distance measures, the generator fails to learn such parameters with the GAN training procedure. We also find that training mixtures of GANs leads to more performance gain compared to increasing the network depth or width when the model complexity is high enough. Our experimental results demonstrate that a mixture of generators can discover different modes or different classes automatically in an unsupervised setting, which we attribute to the distribution of the generation and discrimination tasks across multiple generators and discriminators. As an example of the generalizability of our conclusions to realistic datasets, we train a mixture of GANs on the CIFAR-10 dataset and our method significantly outperforms the state-of-the-art in terms of popular metrics, i.e., Inception Score (IS) and Fréchet Inception Distance (FID).