Conditional Generative Adversarial Nets, Mehdi Mirza, Simon Osindero, 2014arXiv preprint arXiv:1411.1784DOI: 10.48550/arXiv.1411.1784 - Introduces the fundamental concept of Conditional Generative Adversarial Networks, demonstrating how to incorporate class labels into both the generator and discriminator for controlled generation.
Conditional Image Synthesis with Auxiliary Classifier GANs, Augustus Odena, Christopher Olah, Jonathon Shlens, 2017International Conference on Machine Learning (ICML)DOI: 10.48550/arXiv.1610.09585 - Proposes the Auxiliary Classifier GAN architecture, where the discriminator predicts both real/fake and the class label, enhancing the generator's ability to produce class-conditional images.
Large-scale GAN training for high-fidelity natural image synthesis, Andrew Brock, Jeff Donahue, Karen Simonyan, 2019International Conference on Learning Representations (ICLR)DOI: 10.48550/arXiv.1809.11096 - Introduces the Projection Discriminator, significantly improving conditional GAN performance by incorporating an explicit inner product term between image features and class embeddings in the discriminator's output.