Disconnected Manifold Learning for Generative Adversarial Networks

Next, we show how using a collection of generators can address this problem, providing new insights into the success of such multi-generator GANs.