Noise-Tempered Generative Adversarial Networks

摘要: We present a novel method to stabilize the training of generative adversarial networks. The training stability is often undermined by the limited and low-dimensional support of the …