计算机科学

计算机科学 ›› 2025, Vol. 52 ›› Issue (1): 170-182.doi: 10.11896/jsjkx.231100171

• 数据库&大数据&数据科学 • 上一篇    下一篇

融合时频特征的多粒度时间序列对比学习方法

叶力硕, 何志学   

  1. 中国民航大学计算机科学与技术学院 天津 300300
  • 收稿日期:2023-11-27 修回日期:2024-05-09 出版日期:2025-01-15 发布日期:2025-01-09
  • 通讯作者: 何志学(zxhe@cauc.edu.cn)
  • 作者简介:(yls6603@163.com)
  • 基金资助:
    国家重点研发计划(2021YFB1600502);中央高校基本科研业务费专项(3122019121)

Multi-granularity Time Series Contrastive Learning Method Incorporating Time-Frequency Features

YE Lishuo, HE Zhixue   

  1. School of Computer Science and Technology,Civil Aviation University of China,Tianjin 300300,China
  • Received:2023-11-27 Revised:2024-05-09 Online:2025-01-15 Published:2025-01-09
  • About author:YE Lishuo,born in 2001,postgraduate.His main research interests include data mining and time series analysis.
    HE Zhixue,born in 1982,Ph.D,asso-ciate professor,is a member of CCF(No.E2393M).His main research interests include big data processing and analysis,data mining.
  • Supported by:
    National Key Research and Development Program of China(2021YFB1600502)and Fundamental Research Funds for the Central Universities of Ministry of Education of China(3122019121).

PDF (PC)

166

摘要： 现有的时间序列对比学习方法存在增强样本构造方式过于依赖人工经验、泛化能力不足、正样本的定义方式不够通用、对比度量方式存在粗粒度表征等问题,使得整体的时序表示效果较差。为此,提出了一种融合时频特征的多粒度时间序列对比学习方法(Temporal-Spectral Deep Contrastive Network,TSDC)。该方法通过季节-趋势生成网络在时域内产生具有稳定变化的时序增强样本,通过多频带融合扰动操作在频域内产生非稳定变化的时序增强样本,两种增强样本通过实例级别的粗粒度对比以及维度级别的细粒度对比方式进行对比学习,使得模型在获得较好表征的同时能够较好地适应于下游不同类型的时序任务。在多个时间序列公开数据集上进行的分类、预测以及异常检测实验表明,由TSDC方法所得的表征用于下游任务的结果优于典型基线模型。

关键词: 时间序列, 表示学习, 对比学习, 数据增强, 多粒度对比

Abstract: Existing time series contrastive learning methods have some problems,such as augmented sample construction methods rely too much on manual experience,insufficient generalization ability,positive samples are not defined in a general enough way,and coarse-grained representations of contrastive measures,resulting in weak overall time series representation.Therefore,a multi-granularity time series contrastive learning method based on time-frequency features(TSDC) is proposed.The seasonal-trend generation network generates temporal augmentation samples with stable variations in the time domain,and the multi-band fusion perturbation operation generates non-stable variations temporal augmentation samples in the frequency domain,and the two augmentation samples are learned through coarse-grained contrastive at the instance level and fine-grained contrastive at the dimension level,so that the model can be better adapted to different types of downstream time series tasks while obtaining better representation.Experiments on classification,prediction,and anomaly detection on multiple time series public datasets show that the representation obtained by the TSDC method outperforms typical baseline models for downstream tasks.

Key words: Time series, Representation learning, Contrastive learning, Data augmentation, Multi-granularity contrastive

中图分类号: 

  • TP391
[1]LIU H,YIN X,CHEN J,et al.Multi-Dimensional Time-Series Prediction Based on Hybrid Network Models[J].Computer Engineering,2023,49(1):121-129.
[2]LI X,LIU X P,LI W C,et al.Survey on Contrastive Learning Research.Journal of Chinese Computer Systems.2023,44(4):787-797.
[3]YAO X,GAO J Y,XU C S.Self-supervised Graph Contrastive Learning for Video Question Answering[J].Journal of Software,2023,34(5):2083-2100.
[4]YANG B,LUO J C,SONG Y T,et al.Time Series Clustering Method Based on Contrast-ive Learning[J].Computer Science,2024,51(2):63-72.
[5]FRANCESCHI J Y,DIEULEVEUT A,JAGGI M.Unsuper-vised scalable representation learning for multivariate time series[C]//Proceedings of the 33rd International Conference on Neural Information Processing Systems.2019:4650-4661.
[6]TONEKABONI S,EYTAN D,GOLDENBERG A.Unsuper-vised representation learning for time series with temporal neighborhood coding[C]//Proceedings of the 2020 InternationalConference on Learning Representations,2020.
[7]ELDELE E,RAGAB M,CHEN Z,et al.Time-series representation learning via temporal and contextual contrasting [EB/OL].(2021-06-26) [2023-09-15].https://arxiv.org/abs/2106.14112.
[8]YUE Z,WANG Y,DUAN J,et al.Ts2vec:Towards universal representation of time series[C]//Proceedings of the 36th AAAI Conference on Artificial Intelligence.Palo Alto:AAAI,2022:8980-8987.
[9]ZHENG X,CHEN X,SCHURCH M,et al.SimTS:Rethinking Contrastive Representatio-n Learning for Time Series Forecasting.[EB/OL].(2023-03-21) [2023-09-15].https://arxiv.org/abs/2303.18205.
[10]WEN Q,GAO J,SONG X,et al.RobustSTL:A robust seasonal-trend decomposition algorithm for long time series[C]//Proceedings of the 33rd AAAI Conference on Artificial Intelligence.Palo Alto:AAAI,2019:5409-5416.
[11]CLEVELAND R B,CLEVELAND W S,MCRAE J E,et al.STL:A seasonal-trend decomposition[J].J.Off.Stat,1990,6(1):3-73.
[12]ZHANG K,WEN Q,ZHANG C,et al.Self-Supervised Learning for Time Series Analysis:Taxonomy,Progress,and Prospects[J].arXiv:2306.10125,2023.
[13]DOKUMENTOV A,HYNDMAN R J.STR:A seasonal-trend decomposition procedure based on regression[J].arXiv:2009.05894,2020.
[14]WU H X,HU T G ,LIU Y,et al.TimesNet:Temporal 2D-Varia-tion Modeling for General Time Series Analysis[C]//Procee-dings of the 11th International Conference on Learning Representations.2022.
[15]JAISWAL A,BABU A R,ZADEH M Z,et al.A survey on con-trastive self-supervised learning[J].Technologies,2020,9(1):2.
[16]CHEN T,KORNBLITH S,NOROUZI M,et al.A simpleframework for contrastive learning of visual representations[C]//Proceedings of the 2020 International Conference on Machine Learning.PMLR,2020:1597-1607.
[17]OORD A,LI Y,VINYALS O.Representation learning with contrastive predictive coding [EB/OL].(2019-01-22) [2023-09-15].https://arxiv.org/abs/1807.03748.
[18]DAU H A,BAGNALL A,KAMGAR K,et al.The UCR time series archive[J].IEEE/CAA Journal of Automatica Sinica,2019,6(6):1293-1305.
[19]BAGNALL A,DAU H A,LINES J,et al.The UEA multiva-riate time series classification archive [EB/OL].(2018-10-31) [2023-09-15].https://arxiv.org/abs/1811.00075.
[20]WU H,XU J,WANG J,et al.Autoformer:Decomposition transformers with auto-correlation for long-term series forecasting[J].Advances in Neural Information Processing Systems,2021,34:22419-22430.
[21]ABDULAAL A,LIU Z,LANCEWICKI T.Practical approach to asynchronous multivariate time series anomaly detection and localization[C]//Proceedings of the 27th ACM SIGKDD Confe-rence on Knowledge Discovery & Data Mining.New York:ACM,2021:2485-2494.
[22]SU Y,ZHAO Y J,NIU C H,et al.Robust anomaly detection for multivariate time series through stochastic recurrent neural network[C]//Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.New York:ACM,2019:2828-2837.
[23]LI D,CHEN D,JIN B,et al.MAD-GAN:multivariate anomaly detection for time series data with generative adversarial networks[C]//Proceedings of the 2019 International Conference on Artificial Neural Networks.Berlin:Springer,2019:214-232.
[24]ZHAO Y,NASRULLAH Z,LI Z.PyOD:A Python Toolbox for Scalable Outlier Detection[J].Journal of Machine Learning Research,2019,20:1-7.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发