AGAN: Towards Automated Design of Generative Adversarial Networks

Abstract

Recent progress in Generative Adversarial Networks (GANs) has shown promising signs of improving GAN training via architectural change. Despite some early success, at present the design of GAN architectures requires human expertise, laborious trial-and-error testings, and often draws inspiration from its image classification counterpart. In the current paper, we present the first neural architecture search algorithm, automated neural architecture search for deep generative models, or AGAN for abbreviation, that is specifically suited for GAN training. For unsupervised image generation tasks on CIFAR-10, our algorithm finds architecture that outperforms state-of-the-art models under same regularization techniques. For supervised tasks, the automatically searched architectures also achieve highly competitive performance, outperforming best human-invented architectures at resolution $32\times 32$. Moreover, we empirically demonstrate that the modules learned by AGAN are transferable to other image generation tasks such as STL-10.

References in corpus (13)

• Conditional Generative Adversarial Nets
• Neural Architecture Search with Reinforcement Learning
• Large Scale GAN Training for High Fidelity Natural Image Synthesis
• Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
• Improved Training of Wasserstein GANs
• Improved Techniques for Training GANs
• Efficient Neural Architecture Search via Parameter Sharing
• Efficient Architecture Search by Network Transformation
• ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware
• Progressive Neural Architecture Search
• Path-Level Network Transformation for Efficient Architecture Search
• Stacked Generative Adversarial Networks
• Class-Splitting Generative Adversarial Networks

Cited by in corpus (7)

• Auto-GNN: Neural Architecture Search of Graph Neural Networks
• Weight-Sharing Neural Architecture Search: A Battle to Shrink the Optimization Gap
• AdversarialNAS: Adversarial Neural Architecture Search for GANs
• Self-supervised Neural Architecture Search
• AlphaGAN: Fully Differentiable Architecture Search for Generative Adversarial Networks
• Combining Optimal Path Search With Task-Dependent Learning in a Neural Network
• GMM-Based Generative Adversarial Encoder Learning