Collaborative Sampling in Generative Adversarial Networks
This addresses a specific bottleneck in GANs for researchers and practitioners by enhancing sampling efficiency, though it is incremental as it builds on existing GAN frameworks.
The paper tackles the problem of standard GAN sampling discarding the discriminator's learned information, proposing a collaborative sampling scheme that uses the discriminator to refine generated samples through gradient-based updates, resulting in improved sample quality both quantitatively and qualitatively.
The standard practice in Generative Adversarial Networks (GANs) discards the discriminator during sampling. However, this sampling method loses valuable information learned by the discriminator regarding the data distribution. In this work, we propose a collaborative sampling scheme between the generator and the discriminator for improved data generation. Guided by the discriminator, our approach refines the generated samples through gradient-based updates at a particular layer of the generator, shifting the generator distribution closer to the real data distribution. Additionally, we present a practical discriminator shaping method that can smoothen the loss landscape provided by the discriminator for effective sample refinement. Through extensive experiments on synthetic and image datasets, we demonstrate that our proposed method can improve generated samples both quantitatively and qualitatively, offering a new degree of freedom in GAN sampling.