ChainGAN: A sequential approach to GANs
This work addresses the challenge of improving sample quality in generative models for machine learning applications, but it is incremental as it builds on existing GAN frameworks.
The paper tackles the problem of generating data samples by proposing ChainGAN, a two-step GAN architecture that first creates a crude sample and then sequentially refines it with independent editors, achieving competitive results on multiple datasets.
We propose a new architecture and training methodology for generative adversarial networks. Current approaches attempt to learn the transformation from a noise sample to a generated data sample in one shot. Our proposed generator architecture, called $\textit{ChainGAN}$, uses a two-step process. It first attempts to transform a noise vector into a crude sample, similar to a traditional generator. Next, a chain of networks, called $\textit{editors}$, attempt to sequentially enhance this sample. We train each of these units independently, instead of with end-to-end backpropagation on the entire chain. Our model is robust, efficient, and flexible as we can apply it to various network architectures. We provide rationale for our choices and experimentally evaluate our model, achieving competitive results on several datasets.