Perturbative GAN: GAN with Perturbation Layers
This work addresses efficiency and performance issues in GAN training for image generation tasks, but it is incremental as it builds on existing convolutional GANs like DCGAN and WGAN-GP.
The paper tackles the problem of reducing training costs and improving performance in generative adversarial networks (GANs) by proposing Perturbative GAN, which replaces convolution layers with perturbation layers that add fixed noise masks. The result includes a smaller number of parameters, faster convergence, higher inception scores on datasets like CIFAR10 and ImageNet, and reduced overall training cost.
Perturbative GAN, which replaces convolution layers of existing convolutional GANs (DCGAN, WGAN-GP, BIGGAN, etc.) with perturbation layers that adds a fixed noise mask, is proposed. Compared with the convolu-tional GANs, the number of parameters to be trained is smaller, the convergence of training is faster, the incep-tion score of generated images is higher, and the overall training cost is reduced. Algorithmic generation of the noise masks is also proposed, with which the training, as well as the generation, can be boosted with hardware acceleration. Perturbative GAN is evaluated using con-ventional datasets (CIFAR10, LSUN, ImageNet), both in the cases when a perturbation layer is adopted only for Generators and when it is introduced to both Generator and Discriminator.