S2cGAN: Semi-Supervised Training of Conditional GANs with Fewer Labels
This work addresses the cost and scarcity of labeled data for conditional generative models, offering a practical improvement for applications requiring controlled generation.
The paper tackles the problem of expensive labeled data requirements for training conditional GANs by proposing a semi-supervised framework that uses sparse labels for conditional mapping and leverages unsupervised data for distribution learning, demonstrating effectiveness across multiple datasets and tasks.
Generative adversarial networks (GANs) have been remarkably successful in learning complex high dimensional real word distributions and generating realistic samples. However, they provide limited control over the generation process. Conditional GANs (cGANs) provide a mechanism to control the generation process by conditioning the output on a user defined input. Although training GANs requires only unsupervised data, training cGANs requires labelled data which can be very expensive to obtain. We propose a framework for semi-supervised training of cGANs which utilizes sparse labels to learn the conditional mapping, and at the same time leverages a large amount of unsupervised data to learn the unconditional distribution. We demonstrate effectiveness of our method on multiple datasets and different conditional tasks.