Journal
PATTERN RECOGNITION LETTERS
Volume 138, Issue -, Pages 60-67Publisher
ELSEVIER
DOI: 10.1016/j.patrec.2020.06.022
Keywords
Time series segmentation; Multivariate data; Memetic algorithm
Categories
Funding
- Energy Technology Development Business Program of the Korea Institute of Energy Technology Evaluation and Planning (KETEP)
- Ministry of Trade, Industry & Energy, Republic of Korea [20181110100420]
- Korea Institute of Energy Technology Evaluation & Planning (KETEP) [20181110100420] Funding Source: Korea Institute of Science & Technology Information (KISTI), National Science & Technology Information Service (NTIS)
Ask authors/readers for more resources
In recent years, analyzing time-series data has become an ever important research topic due to an increased number of temporal datasets in science and engineering. Segmentation is widely used in time-series data analysis because it provides a more compact representation by dividing the series into segments. Segmentation approaches based on genetic algorithms have been proposed to extract segments and patterns with a given objective, such as a low rate of change or periodicity from time-series data. However, they may not be effective in obtaining the precise solution because they perform global search. In this study, we propose a memetic algorithm for multivariate time-series segmentation. For efficient local refinement, we calculate a likelihood-based score for all time points and use it in the evolutionary process. Experiments demonstrate that the proposed method is superior to conventional segmentation methods. The source code of the proposed method can be downloaded from https://github.com/hlim-kist/ma_mts (C) 2020 Elsevier B.V. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available