Inception Score, Label Smoothing, Gradient Vanishing and -log(D(x)) Alternative
This is an incremental theoretical study that provides mathematical insights into existing GAN issues, primarily relevant for researchers in generative models.
The paper mathematically analyzes several GAN-related topics, such as Inception Score, label smoothing, gradient vanishing, and the -log(D(x)) alternative, but does not present new experimental results or concrete numerical findings.
In this article, we mathematically study several GAN related topics, including Inception score, label smoothing, gradient vanishing and the -log(D(x)) alternative. --- An advanced version is included in arXiv:1703.02000 "Activation Maximization Generative Adversarial Nets". Please refer Section 6 in 1703.02000 for detailed analysis on Inception Score, and refer its appendix for the discussions on Label Smoothing, Gradient Vanishing and -log(D(x)) Alternative.