Research Article
A Dynamic Programming Approach for Time Series Discord Detection
@INPROCEEDINGS{10.1007/978-3-030-93179-7_20, author={Duong Tuan Anh and Nguyen Van Hien}, title={A Dynamic Programming Approach for Time Series Discord Detection}, proceedings={Context-Aware Systems and Applications. 10th EAI International Conference, ICCASA 2021, Virtual Event, October 28--29, 2021, Proceedings}, proceedings_a={ICCASA}, year={2022}, month={1}, keywords={Time series Discord Discord detection Dynamic programming}, doi={10.1007/978-3-030-93179-7_20} }- Duong Tuan Anh
Nguyen Van Hien
Year: 2022
A Dynamic Programming Approach for Time Series Discord Detection
ICCASA
Springer
DOI: 10.1007/978-3-030-93179-7_20
Abstract
There have been several methods to search the top anomaly subsequence (1-discord) in a time series. Most of these methods belong to the window-based category which uses a sliding window with a pre-specified length to extract subsequences. However, one of the main shortcomings of these window-based methods for discord detection is that their computational cost is still high in the cases of very large time series data. In this paper, we propose a new dynamic programming approach for discord detection in time series under Euclidean distance in order to improve further its time efficiency. We evaluate our proposed dynamic programming approach on several time series datasets and the results show that our method provides performance up to 25.2 times faster than HOT SAX algorithm.
