Transitional SAX Representation for Knowledge Discovery for Time Series

被引:7
作者
Song, Kiburm [1 ,2 ]
Ryu, Minho [2 ,3 ]
Lee, Kichun [2 ]
机构
[1] Samsung Electromech, MIS Grp, Global Technol Ctr, Suwon 16674, Gyeonggi Do, South Korea
[2] Hanyang Univ, Dept Ind Engn, Seoul 04763, South Korea
[3] SK Telecom, Vis AI Labs, Seoul 04539, South Korea
来源
APPLIED SCIENCES-BASEL | 2020年 / 10卷 / 19期
基金
新加坡国家研究基金会;
关键词
dimensionality reduction; time-series representation; symbolic aggregate approximation; transition information; SYMBOLIC AGGREGATE APPROXIMATION;
D O I
10.3390/app10196980
中图分类号
O6 [化学];
学科分类号
0703 ;
摘要
Numerous dimensionality-reducing representations of time series have been proposed in data mining and have proved to be useful, especially in handling a high volume of time series data. Among them, widely used symbolic representations such as symbolic aggregate approximation and piecewise aggregate approximation focus on information of local averages of time series. To compensate for such methods, several attempts were made to include trend information. However, the included trend information is quite simple, leading to great information loss. Such information is hardly extendable, so adjusting the level of simplicity to a higher complexity is difficult. In this paper, we propose a new symbolic representation method called transitional symbolic aggregate approximation that incorporates transitional information into symbolic aggregate approximations. We show that the proposed method, satisfying a lower bound of the Euclidean distance, is able to preserve meaningful information, including dynamic trend transitions in segmented time series, while still reducing dimensionality. We also show that this method is advantageous from theoretical aspects of interpretability, and practical and superior in terms of time-series classification tasks when compared with existing symbolic representation methods.
引用
收藏
页码:1 / 14
页数:14
相关论文
共 20 条
  • [1] AGRAWAL R, 1993, 4 INT C FDN DAT ORG, P69
  • [2] [Anonymous], 2003, DMKD, DOI DOI 10.1145/882082.882086
  • [3] Barnaghi P.M., P 2012 IEEE SENSORS
  • [4] SAX Discretization Does Not Guarantee Equiprobable Symbols
    Butler, Matthew
    Kazakov, Dimitar
    [J]. IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, 2015, 27 (04) : 1162 - 1166
  • [5] Locally adaptive dimensionality reduction for indexing large time series databases
    Chakrabarti, K
    Keogh, E
    Mehrotra, S
    Pazzani, M
    [J]. ACM TRANSACTIONS ON DATABASE SYSTEMS, 2002, 27 (02): : 188 - 228
  • [6] Efficient time series matching by wavelets
    Chan, KP
    Fu, AWC
    [J]. 15TH INTERNATIONAL CONFERENCE ON DATA ENGINEERING, PROCEEDINGS, 1999, : 126 - 133
  • [7] Chen Y., 2015, The UCR Time Series Classification Archive
  • [8] Fuad Muhammad Marwan Muhammad, 2012, Data Warehousing and Knowledge Discovery. Proceedings of the 14th International Conference, DaWaK 2012, P105, DOI 10.1007/978-3-642-32584-7_9
  • [9] Dimensionality Reduction for Fast Similarity Search in Large Time Series Databases
    Eamonn Keogh
    Kaushik Chakrabarti
    Michael Pazzani
    Sharad Mehrotra
    [J]. Knowledge and Information Systems, 2001, 3 (3) : 263 - 286
  • [10] Korn F., 1997, SIGMOD Record, V26, P289, DOI 10.1145/253262.253332