基于多鉴别器生成对抗网络的时间序列生成模型

doi:10.11959/j.issn.1000-436x.2022205

通信学报 ›› 2022, Vol. 43 ›› Issue (10): 167-176.doi: 10.11959/j.issn.1000-436x.2022205

基于多鉴别器生成对抗网络的时间序列生成模型

陆彦辉¹^,², 柳寒¹^,², 李航², 朱光旭²

¹ 郑州大学电气与信息工程学院，河南郑州 450001
² 深圳市大数据研究院，广东深圳 518115

修回日期:2022-10-09 出版日期:2022-10-25 发布日期:2022-10-01
作者简介:陆彦辉（1972− ），女，河南许昌人，郑州大学教授，主要研究方向为宽带无线通信理论与系统、无线资源管理和机器学习等
柳寒（1997− ），男，河南邓州人，郑州大学硕士生，主要研究方向为人工智能与大数据处理、大数据分析与数据挖掘
李航（1985− ），男，河北承德人，深圳市大数据研究院副研究员，主要研究方向为无线通信与网络、机器学习等
朱光旭（1989− ），男，广东广州人，深圳市大数据研究院副研究员，主要研究方向为边缘智能、联邦学习、通信感知一体化等
基金资助:
国家自然科学基金资助项目(62001310);广东省基础与应用基础研究基金资助项目(2022A1515010109);深圳市科技计划基础研究项目(JCYJ20190813170803617)

Time series generation model based on multi-discriminator generative adversarial network

Yanhui LU¹^,², Han LIU¹^,², Hang LI², Guangxu ZHU²

¹ College of Electrical and Information Engineering, Zhengzhou University, Zhengzhou 450001, China
² Shenzhen Research Institute of Big Data, Shenzhen 518115, China

Revised:2022-10-09 Online:2022-10-25 Published:2022-10-01
Supported by:
The National Natural Science Foundation of China(62001310);The Foundation for Basic and Applied Basic Research of Guangdong Province(2022A1515010109);The Basic Research Project of Shenzhen Science and Technology Plan(JCYJ20190813170803617)

摘要/Abstract

摘要：

摘要：针对时间序列的隐私性和连续性导致时间序列数据集在收集过程中存在收集代价昂贵和数据缺失等问题，提出了一种基于循环神经网络的多鉴别器生成对抗网络模型，该模型能够利用小规模数据集合成得到与真实数据相似分布的时间序列数据集。多鉴别器包含时域、频域、时频域和自相关4种鉴别器，能够充分识别时间序列不同维度下的特征。在实验中，通过损失函数的收敛分析、主成分分析和误差分析，分别从定性和定量的角度对模型进行性能评估。结果表明，所提模型和其他参考模型相比具有更好的性能。

关键词: 生成对抗网络, 时间序列, 傅里叶变换, 自相关函数, 机器学习

Abstract:

Aiming at the problems of expensive collection cost and missing data due to the privacy and continuity of time series data set, a multi-discriminator generative adversarial network model based on recurrent neural network was proposed, which could synthesize time series dataset that were approximately distributed with real data of a small scale dataset.Multi-discriminator included four discriminators in time domain, frequency domain, time-frequency domain and autocorrelation.Different discriminators could effectively recognize the features of the time series in different domains.In the experiment, the convergence of loss function, principal component analysis and error analysis were performed to evaluate the performance of the model from qualitative and quantitative perspectives.The experimental results show that the proposed model has better performance than other reference models.

Key words: generative adversarial network, time series, Fourier transform, autocorrelation function, machine learning

中图分类号:

TP18

陆彦辉, 柳寒, 李航, 朱光旭. 基于多鉴别器生成对抗网络的时间序列生成模型[J]. 通信学报, 2022, 43(10): 167-176.

Yanhui LU, Han LIU, Hang LI, Guangxu ZHU. Time series generation model based on multi-discriminator generative adversarial network[J]. Journal on Communications, 2022, 43(10): 167-176.

图/表 11

图1

图2

图3

图4

图5

表1

图6

表2

图7

图8

表3

参考文献 24

[1]	NAEEM M , JAMAL T , DIAZ-MARTINEZ J ,et al. Trends and future perspective challenges in big data[C]// Advances in Intelligent Data Analysis and Applications. Berlin:Springer, 2022: 309-325.
[2]	HE H T , GAO S C , JIN T ,et al. A seasonal-trend decomposition-based dendritic neuron model for financial time series prediction[J]. Applied Soft Computing, 2021,108:107488.
[3]	ZHANG M , LIN H , LONG X R ,et al. Analyzing the spatiotemporal pattern and driving factors of wetland vegetation changes using 2000—2019 time-series Landsat data[J]. Science of the Total Environment, 2021,780:146615.
[4]	AL-SHABANDAR R , JADDOA A , LIATSIS P ,et al. A deep gated recurrent neural network for petroleum production forecasting[J]. Machine Learning With Applications, 2021,3:100013.
[5]	BALLI S . Data analysis of Covid-19 pandemic and short-term cumulative case forecasting using machine learning time series methods[J]. Chaos,Solitons ＆ Fractals, 2021,142:110512.
[6]	THAPA C , CAMTEPE S . Precision health data:requirements,challenges and existing techniques for data security and privacy[J]. Computers in Biology and Medicine, 2021,129:104130.
[7]	HUANG T G , CHAKRABORTY P , SHARMA A . Deep convolutional generative adversarial networks for traffic data imputation encoding time series as images[J]. International Journal of Transportation Science and Technology, 2021:doi.org/10.1016/j.ijtst.2021.10.007.
[8]	IWANA B K , UCHIDA S . An empirical survey of data augmentation for time series classification with neural networks[J]. PLoSOne, 2021,16(7): e0254841.
[9]	KINGMA D P , WELLING M . Auto-encoding variational Bayes[J]. arXiv Preprint,arXiv:1312.6114, 2013.
[10]	GOODFELLOW I , POUGET-ABADIE J , MIRZA M ,et al. Generative adversarial networks[J]. Communications of the ACM, 2020,63(11): 139-144.
[11]	RADFORD A , METZ L , CHINTALA S . Unsupervised representation learning with deep convolutional generative adversarial networks[J]. arXiv Preprint,arXiv:1511.06434, 2015.
[12]	ARJOVSKY M , CHINTALA S , BOTTOU L . Wasserstein generative adversarial networks[C]// Proceedings of the 34th International Conference on Machine Learning. [S.l.]:JMLR, 2017: 214-223.
[13]	ISOLA P , ZHU J Y , ZHOU T H ,et al. Image-to-image translation with conditional adversarial networks[C]// Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway:IEEE Press, 2017: 5967-5976.
[14]	ZHU J Y , PARK T , ISOLA P ,et al. Unpaired image-to-image translation using cycle-consistent adversarial networks[C]// Proceedings of 2017 IEEE International Conference on Computer Vision. Piscataway:IEEE Press, 2017: 2242-2251.
[15]	KARRAS T , LAINE S , AILA T M . A style-based generator architecture for generative adversarial networks[C]// Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway:IEEE Press, 2019: 4396-4405.
[16]	MEDSKER L R , JAIN L C . Recurrent neural networks[J]. Design and Applications, 2001,5: 64-67.
[17]	GRAVES A . Long short-term memory[J]. Supervised Sequence Labelling with Recurrent Neural Networks, 2012,385: 37-45.
[18]	MOGREN O . C-RNN-GAN:continuous recurrent neural networks with adversarial training[J]. arXiv Preprint,arXiv:1611.09904, 2016.
[19]	ESTEBAN C , HYLAND S L , R?TSCH G , . Real-valued (medical) time series generation with recurrent conditional GANs[J]. arXiv Preprint,arXiv:1706.02633, 2017.
[20]	YOON J , JARRETT D , VAN DER SCHAAR M . Time-series generative adversarial networks[C]// Proceedings of the 33rd Conference on Neural Information Processing Systems (NeurIPS 2019).[S.l.:s.n.], 2019: 1-11.
[21]	NI H , SZPRUCH L , WIESE M ,et al. Conditional Sig-Wasserstein GANs for time series generation[J]. SSRN Electronic Journal, 2020:doi.org/10.2139/ssrn.3623086.
[22]	OPPENHEIM A V , WILLSKY A S , NAWAB S H ,et al. Signals ＆systems[M]. New York: Pearson Educación, 1997.
[23]	BAROT T , BURGSTEINER H , KOLLERITSCH W . Comparison of discrete autocorrelation functions with regards to statistical significance[C]// Applied Informatics and Cybernetics in Intelligent Systems. Berlin:Springer, 2020: 257-266.
[24]	KINGMA D P , BA J . Adam:a method for stochastic optimization[J]. arXiv Preprint,arXiv:1412.6980, 2014.

训练模型	MSE	RMSE	MAE	MAPE
MDGAN	1.175 438	1.082 374	0.863 169	7.125 314
频域鉴别器GAN	1.330 979	1.180 982	1.147 531	8.293 083
时频域鉴别器GAN	1.293 977	1.271 668	1.136 859	8.570 859
自相关鉴别器GAN	1.411 853	1.310 379	1.362 662	9.135 859

训练模型	MSE	RMSE	MAE	MAPE
MDGAN	1.175 438	1.082 374	0.844 753	7.047 909
TimeGAN	1.182 172	1.106 035	0.863 169	7.125 314
SigCWGAN	1.241 732	1.111 821	0.907 567	7.405 342
RCGAN	1.388 121	1.234 397	1.021 464	8.490 861

训练模型	MSE	RMSE	MAE	MAPE
MDGAN	0.002 614	0.035 102	0.028 508	11.761 79
TimeGAN	0.002 981	0.051 701	0.041 101	20.810 77
SigCWGAN	0.003 176	0.050 541	0.041 165	20.214 82
RCGAN	0.004 475	0.066 354	0.054 091	28.773 71

基于多鉴别器生成对抗网络的时间序列生成模型

Time series generation model based on multi-discriminator generative adversarial network

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 24

相关文章 15

Metrics

推荐阅读 0

[1]	张佳乐, 朱诚诚, 孙小兵, 陈兵. 基于GAN的联邦学习成员推理攻击与防御方法[J]. 通信学报, 2023, 44(5): 193-205.
[2]	苏新, 张桂福, 行鸿彦, Zenghui Wang. 基于平衡生成对抗网络的海洋气象传感网入侵检测研究[J]. 通信学报, 2023, 44(4): 124-136.
[3]	戴千一, 张斌, 郭松, 徐开勇. 基于多分类器集成的区块链网络层异常流量检测方法[J]. 通信学报, 2023, 44(3): 66-80.
[4]	霍纬纲, 梁锐, 李永华. 基于随机Transformer的多维时间序列异常检测模型[J]. 通信学报, 2023, 44(2): 94-103.
[5]	刘延华, 李嘉琪, 欧振贵, 高晓玲, 刘西蒙, MENG Weizhi, 刘宝旭. 对抗训练驱动的恶意代码检测增强方法[J]. 通信学报, 2022, 43(9): 169-180.
[6]	王延文, 雷为民, 张伟, 孟欢, 陈新怡, 叶文慧, 景庆阳. 基于生成模型的视频图像重建方法综述[J]. 通信学报, 2022, 43(9): 194-208.
[7]	李昂, 陈建新, 魏昕, 周亮. 面向6G的跨模态信号重建技术[J]. 通信学报, 2022, 43(6): 28-40.
[8]	段雪源, 付钰, 王坤. 基于VAE-WGAN的多维时间序列异常检测方法[J]. 通信学报, 2022, 43(3): 1-13.
[9]	向夏雨, 王佳慧, 王子睿, 段少明, 潘鹤中, 庄荣飞, 韩培义, 刘川意. 基于生成对抗网络技术的医疗仿真数据生成方法[J]. 通信学报, 2022, 43(3): 211-224.
[10]	何高峰, 魏千峰, 肖咸财, 朱海婷, 徐丙凤. 支持数据隐私保护的恶意加密流量检测确认方法[J]. 通信学报, 2022, 43(2): 156-170.
[11]	冯智斌, 徐煜华, 杜智勇, 刘鑫, 李文, 韩昊, 张晓博. 对抗智能干扰的主动防御技术[J]. 通信学报, 2022, 43(10): 42-54.
[12]	刘威, 陈成, 江锐, 卢涛. 四通道无监督学习图像去雾网络[J]. 通信学报, 2022, 43(10): 210-222.
[13]	梅锴, 赵海涛, 刘潇然, 刘军, 熊俊, 任保全, 魏急波. 高效的基于数据与模型的信道估计算法[J]. 通信学报, 2022, 43(1): 59-70.
[14]	彭长根, 高婷, 刘惠篮, 丁红发. 面向机器学习模型的基于PCA的成员推理攻击[J]. 通信学报, 2022, 43(1): 149-160.
[15]	周志立, 王美民, 杨高波, 朱剑宇, 孙星明. 基于轮廓自动生成的构造式图像隐写方法[J]. 通信学报, 2021, 42(9): 144-154.