Abstract
The challenges in effcient data representation and similarity measures on massive amounts of time series have enormous impact on many applications. This paper addresses an improvement on Symbolic Aggregate approXimation (SAX), is one of the effcient representations for time series mining. Because SAX represents its symbols by the average (mean) value of a segment with the assumption of Gaussian distribution, it is insuficient to serve the entire deterministic information and causes sometimes incorrect results in time series classiffcation. In this work, SAX representation and distance measure is improved with the addition of another moment of the prior distribution, standard deviation; SAX SD is proposed. We provide comprehensive analysis for the proposed SAX SD and conrm both the highest classi-fication accuracy and the highest dimensionality reduction ratio on University of California, Riverside (UCR) datasets in comparison to state of the art methods such as SAX, Extended SAX (ESAX) and SAX Trend Distance (SAX TD).
| Original language | English |
|---|---|
| Title of host publication | 18th International Conference on Information Integration and Web-Based Applications and Services, iiWAS 2016 - Proceedings |
| Publisher | Association for Computing Machinery |
| Pages | 72-80 |
| Number of pages | 9 |
| Volume | Part F126325 |
| ISBN (Electronic) | 9781450348072 |
| DOIs | |
| Publication status | Published - 2016 Nov 28 |
| Event | 18th International Conference on Information Integration and Web-Based Applications and Services, iiWAS 2016 - Singapore, Singapore Duration: 2016 Nov 28 → 2016 Nov 30 |
Other
| Other | 18th International Conference on Information Integration and Web-Based Applications and Services, iiWAS 2016 |
|---|---|
| Country | Singapore |
| City | Singapore |
| Period | 16/11/28 → 16/11/30 |
Fingerprint
Keywords
- Classi-cation
- Dimension reduction
- Statistical features
- Symbolic representation
- Time series
ASJC Scopus subject areas
- Human-Computer Interaction
- Computer Networks and Communications
- Computer Vision and Pattern Recognition
- Software
Cite this
An improved symbolic aggregate approximation distance measure based on its statistical features. / Thet Zan, Chaw; Yamana, Hayato.
18th International Conference on Information Integration and Web-Based Applications and Services, iiWAS 2016 - Proceedings. Vol. Part F126325 Association for Computing Machinery, 2016. p. 72-80.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - An improved symbolic aggregate approximation distance measure based on its statistical features
AU - Thet Zan, Chaw
AU - Yamana, Hayato
PY - 2016/11/28
Y1 - 2016/11/28
N2 - The challenges in effcient data representation and similarity measures on massive amounts of time series have enormous impact on many applications. This paper addresses an improvement on Symbolic Aggregate approXimation (SAX), is one of the effcient representations for time series mining. Because SAX represents its symbols by the average (mean) value of a segment with the assumption of Gaussian distribution, it is insuficient to serve the entire deterministic information and causes sometimes incorrect results in time series classiffcation. In this work, SAX representation and distance measure is improved with the addition of another moment of the prior distribution, standard deviation; SAX SD is proposed. We provide comprehensive analysis for the proposed SAX SD and conrm both the highest classi-fication accuracy and the highest dimensionality reduction ratio on University of California, Riverside (UCR) datasets in comparison to state of the art methods such as SAX, Extended SAX (ESAX) and SAX Trend Distance (SAX TD).
AB - The challenges in effcient data representation and similarity measures on massive amounts of time series have enormous impact on many applications. This paper addresses an improvement on Symbolic Aggregate approXimation (SAX), is one of the effcient representations for time series mining. Because SAX represents its symbols by the average (mean) value of a segment with the assumption of Gaussian distribution, it is insuficient to serve the entire deterministic information and causes sometimes incorrect results in time series classiffcation. In this work, SAX representation and distance measure is improved with the addition of another moment of the prior distribution, standard deviation; SAX SD is proposed. We provide comprehensive analysis for the proposed SAX SD and conrm both the highest classi-fication accuracy and the highest dimensionality reduction ratio on University of California, Riverside (UCR) datasets in comparison to state of the art methods such as SAX, Extended SAX (ESAX) and SAX Trend Distance (SAX TD).
KW - Classi-cation
KW - Dimension reduction
KW - Statistical features
KW - Symbolic representation
KW - Time series
UR - http://www.scopus.com/inward/record.url?scp=85014891332&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85014891332&partnerID=8YFLogxK
U2 - 10.1145/3011141.3011146
DO - 10.1145/3011141.3011146
M3 - Conference contribution
VL - Part F126325
SP - 72
EP - 80
BT - 18th International Conference on Information Integration and Web-Based Applications and Services, iiWAS 2016 - Proceedings
PB - Association for Computing Machinery
ER -