计算机科学与探索
Journal of Frontiers of Computer Science and Technology doi: 10.3778/j.issn.1673-9418.1909058
* The National Natural Science Foundation of China under Grant No. 61872152 (国家自然科学基金); the National Natural Science Foundation of
China under Grant No. 61872409 (国家自然科学基金); 2018 Provincial Rural Revitalization Strategy Special Project of Guangdong Provincial
Department of Agriculture under Grant No.54 (2018 年广东省农业厅省级乡村振兴战略专项项目(粤农计〔2018〕54 号)); Guangdong
Program for Special Support of Top-notch Young Professionals under Grant No. 2015TQ01X79 (广东省特支计划).
生成对抗网络的研究进展综述
吴少乾,李西明
+
华南农业大学 数学与信息学院,广州 510642
+ 通讯作者 E-mail: liximing@scau.edu.cn
摘 要:自生成对抗网络诞生以来,对其的研究已经成为机器学习领域的一个热点。它利用对抗学习的机制训练模
型,解决了当年生成算法无法解决的问题。由于 GANs 的优势,研究者们对其进行深入地研究,产生了许多 GANs
的衍生模型,这使得 GANs 得到了快速的发展,形成了所谓的 GAN-Zoo。GANs 被广泛应用于视觉领域、音频领域、
自然语言领域及其他各种领域中,如图像生成、图像翻译、文本生成、音频转换和自然语言翻译等。从传统 GANs
出发,对近几年内 GANs 的研究中较为突出的方面进行介绍总结,首先介绍了传统 GANs 的基本理论,然后对近年
来 GANs 的主要衍生模型进行总结,最后总结了 GANs 在图像领域和信息安全领域中的主要应用成果。
关键词:生成对抗网络;散度函数;神经网络;生成模型
文献标志码:A 中图分类号:TP30
吴少乾,李西明. 生成对抗网络的研究进展综述[J].计算机科学与探索
WU S Q, LI X M. A Survey on the Research Progress of Generating Adversarial Networks[J]. Journal of Frontiers of
Computer Science and Technology
A Survey on the Research Progress of Generating Adversarial Networks
WU Shaoqian, LI Ximing
+
College of Mathematics and Informatics, South China Agricultural University, Guangzhou 510642, China
Abstract: Since the birth of Generative Adversarial Networks, the research on it has become a hot spotin the field of
machine learning. It uses the mechanism training model of Adversarial Learning to solve the problem that the generation
algorithm cannot solve. Due to the advantages of GANs, researchers have conducted in-depth research on it and produced a
large number of derivative models of GANs, which empower the rapid development of GANs and the formation of so-called
GAN-Zoo. GANs is widely used in visual field, audio field, natural language field and other fields, such as image generation,
image translation, text generation, audio conversion, natural language translation and so on. Based on the traditional GANs,
we introduce and summarizes the prominent aspects of GANs research in recent years. First, we introduce the basic theory
of GANs, the summarize the main derivative models of GANsin recent years, and finally summarize the main application
results of GANs in image field and information security field.
Key words: GANs; Divergence function; Neural networks; Generative model
评论0