Information-theoretic stochastic contrastive conditional GAN: InfoSCC-GAN

Abstract

Conditional generation is a subclass of generative problems where the output of the generation is conditioned by the attribute information. In this paper, we present a stochastic contrastive conditional generative adversarial network (InfoSCC-GAN) with an explorable latent space. The InfoSCC-GAN architecture is based on an unsupervised contrastive encoder built on the InfoNCE paradigm, an attribute classifier and an EigenGAN generator. We propose a novel training method, based on generator regularization using external or internal attributes every n-th iteration, using a pre-trained contrastive encoder and a pre-trained classifier. The proposed InfoSCC-GAN is derived based on an information-theoretic formulation of mutual information maximization between input data and latent space representation as well as latent space and generated data. Thus, we demonstrate a link between the training objective functions and the above information-theoretic formulation. The experimental results show that InfoSCC-GAN outperforms the "vanilla" Eigen-GAN in the image generation on AFHQ and CelebA datasets. In addition, we investigate the impact of discriminator architectures and loss functions by performing ablation studies. Finally, we demonstrate that thanks to the EigenGAN generator, the proposed framework enjoys a stochastic generation in contrast to vanilla deterministic GANs yet with the independent training of encoder, classifier, and generator in contrast to existing frameworks. Code, experimental results, and demos are available online at github.com/vkinakh/InfoSCC-GAN.

Ensemble WGAN (EWG): Advancing Image Synthesis and Deepfake Detection with Heterogeneous Discriminator Approach

“…On the CelebA dataset, Benny et al likewise attained high accuracy scores.On the CIFAR10 dataset, an accuracy score of 69.51%, and 98.90% on the MNIST dataset was reported by [20], is considerably lower than that of our proposed EWG approach. On the CelebA dataset, [21] received accuracy ratings of 93.27% for baldness, 99.88% for eyeglasses, 95.68% for moustache, 94.62% for wearing a hat, and 98.62% for wearing a necktie. [22] had moderate accuracy scores.…”