生成式对抗网络技术与研究进展

摘要/Abstract

摘要： 近年来，生成对抗网络（generative adversarial networks，GAN）技术研究成果呈指数级增长，生成对抗网络使用零和博弈论将2个相互竞争的神经网络结合起来，使其能够产生更加清晰、更离散的输出．在诸如其计算机视觉、医疗、金融等领域的图像和视频处理及生成、数据集增强、时间序列预测等方面已取得重大进展．本文主要介绍了生成对抗网络的基本框架、理论以及实现过程，并且分析了近年来主流研究现状，通过回顾生成对抗网络变体及其应用场景列举了其亟待改进的问题．此外，本文还着重讨论了如何采用生成对抗网络隐私措施以及处理敏感数据，以及生成对抗网络技术在相关领域的未来发展趋势．

关键词: 深度学习, 生成对抗网络, 数据隐私保护, 数据集增强, 时间序列预测, 差分隐私

Abstract: In recent years, generative adversarial networks (GANs) researches have increased exponentially. The generative adversarial networks utilize zero-sum game theory to combine two competing neural networks, so that they can produce clearer and much more discrete outputs. In the fields of computer vision, medical treatment, finance, etc., significant progress has been made in the field of image and video processing and generation, data set enhancement, and time sequence prediction. This paper introduces the basic framework, theory and implementation process of generative adversarial networks, and analyzes the mainstream research status in recent years, and lists the problems which need to be improved by reviewing the variants of generative adversarial networks and their application scenarios. In addition, this paper also focuses on how to apply generative adversarial networks to arrange privacy measures and deal with sensitive data, as well as the future development trend of generation countermeasure network technology in related fields.

Key words: deep learning, generative adversarial network, data privacy protection, data set enhancement, time sequence prediction, differential privacy

梁晨, 王利斌, 李卓群, 薛源, . 生成式对抗网络技术与研究进展[J]. 信息安全研究, 2022, 8(3): 235-.

参考文献

[1] 王万良, 李卓蓉. 生成式对抗网络研究进展 [J]. 通信学报, 2018, 39(2):135-148

[2] Goodfellow J I, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets [C] //Proc of the Advances in Neural Information Processing Systems. New York: ACM, 2014: 1-9

[3] Nikolaidis K, Kristiansen S, Goebel V, et al. Augmenting physiological time series data: A case study for sleep apnea detection [C] //Proc of the ECML PKDD 2019: Machine Learning and Knowledge Discovery in Databases. Berlin: Springer, 2019: 376-399

[4] Hazra D, Byun Y-C. SynSigGAN: Generative adversarial networks for synthetic biomedical signal generation [J]. Biology, 2020, 9(12): 441:1-441:20

[5] Yoon J, Jarrett D, Schaar M. Time-series generative adversarial networks [C] //Proc of the Advances in Neural Information Processing Systems. New York: ACM, 2019: 5508–5518

[6] Kaushik S, Choudhury A, Natarajan S, et al. Medicine expenditure prediction via a variance-based generative adversarial network [J]. IEEE Access, 2020, 8: 110947-110958

[7] Guo Zijian, Wan Yiming, Ye Hao. A data imputation method for multivariate time series based on generative adversarial network [J]. Neurocomputing, 2019, 360: 185-197

[8] Kolokolova A, Billard M, Bishop R, et al. Gans & reels: Creating irish music using a generative adversarial network [J]. arXiv preprint, arXiv:2010.15772, 2020

[9] Cheng P-S, Lai C-Y, Chang C-C, et al. A variant model of TGAN for music generation [C] //Proc of the 2020 Asia Service Sciences and Software Engineering Conf. New York: ACM, 2020: 40-45

[10] Zhu Guangxuan, Zhao Hongbo, Liu Haoqiang, et al. A novel LSTM-GAN algorithm for time series anomaly detection [C] //Proc of the 2019 Prognostics and System Health Management Conf. Piscataway, NJ: IEEE, 2019: 1-6

[11] Esteban C, Hyland S L, Rätsch G. Real-valued (medical) time series generation with recurrent conditional GANs [J]. arXiv preprint, arXiv:1706.02633, 2017

[12] Culnane C, Rubinstein B I P, Teague V. Health data in an open world [J]. arXiv preprint, arXiv:1712.05627, 2017

[13] Quan H, Dinh N T, Trung L, et al. MGAN: Training generative adversarial nets with multiple generators [C] //Proc of the Int Conf on Learning Representations. BC, Canada: ICLR, 2018:1-23

[14] Benyamin G, Aydin G, Mark C, et al. Fitting a mixture distribution to data: Tutorial [J]. arXiv preprint, arXiv:1901.06708, 2019

[15] Dinh N T, Trung L, Hung V, et al. Dual discriminator generative adversarial nets [C] //Proc of the 31st Int Conf on Neural Information Processing Systems. Cambridge, MA: MIT Press, 2017: 2667-2677

[16] Martin A, Soumith C, Leon B. Wasserstein generative adversarial networks [C] //Proc of the Int Conf on machine learning.New York: ACM, 2017: 214-223

[17] Ishaan G, Faruk A, Martin A, et al. Improved training of wasserstein GANs [C] //Proc of the 31st Int Conf on Neural Information Processing Systems. Cambridge, MA: MIT Press, 2017: 5769-5779

[18] Mirza M, Osindero S. Conditional generative adversarial nets [J]. arXiv preprint, arXiv:1411.1784, 2014: 1-7

[19] Springenberg J T, Dosovitskiy A, Brox T, et al. Striving for simplicity: The all convolutional net [C] //Proc of the Int Conf on Learning Representations. BC, Canada: ICLR, 2015: 1-14

[20] Jordon J, Yoon J, Schaar M. PATE-GAN: Generating synthetic data with differential privacy guarantees [C] //Proc of the Int Conf on Learning Representations. BC, Canada: ICLR, 2019:1-21

[21] Dwork C. Differential privacy [C] //Proc of the 33rd Int Conf on Automata, Languages and Programming-Volume Part II. Berlin: Springer, 2006:1-12

[22] Dwork C, Roth A. The algorithmic foundations of differential privacy [J]. Foundations and Trends^® in Theoretical Computer Science, 2014, 9 (3/4): 211-407

[23] Song J, Ye J C. Federated CycleGAN for privacy-preserving image-to-image translation [J]. arXiv preprint, arXiv: 2106.09246, 2021

[24] Ho S, Qu Y, Gu B, et al. DP-GAN: Differentially private consecutive data publishing using generative adversarial nets [J]. Journal of Network and Computer Applications, 2021, 185: 103066:1-103066:11

[25] Wiese M, Knobloch R, Korn R, et al. Quant gans: Deep generation of financial time series [J]. Quantitative Finance, 2020, 20(9):1419-1440

[26] Qu F, Liu J, Ma Y, et al. A novel wind turbine data imputation method with multiple optimizations based on GANs [J]. Mechanical Systems and Signal Processing, 2020, 139: 106610:1-106610:16

[27] Leangarun T, Tangamchit P, Thajchayapong S. Stock price manipulation detection using generative adversarial networks [C] //Proc of the 2018 IEEE Symp Series on Computational Intelligence. Piscataway, NJ: IEEE, 2018: 2104-2111

[28] 陈阳.著作权法下“换脸技术”的法律约束缺位与规制路径 [J]. 信息安全研究,2020,6(12):1109-1117

[29] 周琳娜,吕欣一.基于GAN图像生成的信息隐藏技术综述 [J]. 信息安全研究,2019,5(9):771-777