Generalization error property of infoGAN for two-layer neural network

Mahmud Hasan, Mathias Muia

Information Maximizing Generative Adversarial Network (infoGAN) can be understood as a minimax problem involving two neural networks: discriminators and generators with mutual info…