人工智能研究的新前线:生成式对抗网络

生成式对抗网络（Generative adversarial networks，GAN）是当前人工智能学界最为重要的研究热点之一.其突出的生成能力不仅可用于生成各类图像和自然语言数据，还启发和推动了各类半监督学习和无监督学习任务的发展.本文概括了GAN的基本思想，并对近年来相关的理论与应用研究进行了梳理，总结了GAN常见的网络结构与训练方法，博弈形式，集成方法，并对一些应用场景进行了介绍.在此基础上，本文对GAN发展的内在逻辑进行了归纳总结.

The New Frontier of AI Research: Generative Adversarial Networks

Recently, generative adversarial networks (GAN) have become one of the most popular topics in artificial intelligent field. Its outstanding capability of generating realistic samples not only revived the research of generative model, but also inspired the research of semi-supervised learning and unsupervised learning. In this paper, we introduce the basic idea of GAN, and comb its recent development in theory and practice. By concluding its improvements of network structures, optimization methods, the form of the game, the ensemble methods, and its applications, we found the inner logic of its development.