Meaningful Subsequence Matching under Time Warping Distance for Data Stream

2009 ◽  
pp. 1013-1020 ◽  
Author(s):  
Vit Niennattrakul ◽  
Chotirat Ann Ratanamahatana
Keyword(s):  
Data Stream ◽  
Time Warping ◽  
Subsequence Matching

Accurate Subsequence Matching on Data Stream under Time Warping Distance

2010 ◽  
pp. 156-167 ◽  
Author(s):  
Vit Niennattrakul ◽  
Dechawut Wanichsan ◽  
Chotirat Ann Ratanamahatana
Keyword(s):  
Data Stream ◽  
Time Warping ◽  
Subsequence Matching

scholarly journals On Extending the Prefix-Querying Method for Efficient Time-Series Subsequence Matching Under Time Warping

2006 ◽  
Vol 13D (3) ◽  
pp. 357-368
Author(s):  
Byoung-Chol Chang ◽  
Sang-Wook Kim ◽  
Jae-Hyuk Cha
Keyword(s):  
Time Series ◽  
Time Warping ◽  
Subsequence Matching

scholarly journals Efficient Subsequence Search on Streaming Data Based on Time Warping Distance

1970 ◽  
Vol 5 (1) ◽  
pp. 2-8 ◽  
Author(s):  
Sura Rodpongpun ◽  
Vit Niennattrakul ◽  
Chotirat Ann Ratanamahatana
Keyword(s):  
Time Series ◽  
Similarity Search ◽  
Time Series Data ◽  
Distance Measure ◽  
Computational Cost ◽  
Streaming Data ◽  
Series Data ◽  
Search Problem ◽  
Time Warping ◽  
Subsequence Matching

Many algorithms have been proposed to deal with subsequence similarity search problem in time series data stream. Dynamic Time Warping (DTW), which has been accepted as the best distance measure in time series similarity search, has been used in many research works. SPRING and its variance were proposed to solve such problem by mitigating the complexity of DTW. Unfortunately, these algorithms produce meaningless result since no normalization is taken into account before the distance calculation. Recently, GPUs and FPGAs were used in similarity search supporting subsequence normalization to reduce the computation complexity, but it is still far from practical use. In this work, we propose a novel Meaningful Subsequence Matching (MSM) algorithm which produces meaningful result in subsequence matching by considering global constraint, uniform scaling, and normalization. Our method significantly outperforms the existing algorithms in terms of both computational cost and accuracy.

Sign in / Sign up

Export Citation Format

Share Document