BriefGPT.xyz
Jun, 2018
对抗网络的超越局部纳什均衡
Beyond Local Nash Equilibria for Adversarial Networks
HTML
PDF
Frans A. Oliehoek, Rahul Savani, Jose Gallego, Elise van der Pol, Roderich Groß
TL;DR
通过将生成对抗网络明确建模为混合策略有限博弈，该论文提出了一种资源有限Nash均衡解决方案，可通过增加计算资源找到更好的解决方案，证明了该方法比GAN和MGAN产生的解决方案不易被操纵，且与理论预测的NEs非常相似。
Abstract
Save for some special cases, current training methods for
generative adversarial networks
(GANs) are at best guaranteed to converge to a `local
nash equilibrium
` (LNE). Such LNEs, however, can be arbitrarily far
→