DAPs

Mining using change-point detection of epileptic activity time series data

Sun Hee Kim, Lei Li, Christos Faloutsos, Hyung Jeong Yang, Seong Whan Lee

Research output: Contribution to journalArticle

Abstract

The goal of this study is to mine meaningful patterns effectively and efficiently via change-point detection of the time series data, with the assistance of domain knowledge and observed data. With those patterns, our method can do segmentation and compression. We developed a novel gray-box approach for mining such data: Domain Assisted Parameter semi-free wave mining (DAPs). DAPs is intended for mining time series with rich domain-specific knowledge based on a chaos model. Specifically, it automatically detects a change-point of time sequences, respecting the minimal description length principle. And the time sequence is segmented based on the detected change-point, and each segment is fitted with a consistent model. The experimental results using both synthetic and real EEG data indicated that the developed method offers a significant improvement in segmentation and compression via pattern detection over other existing methods. DAPs reduced the number of bits of the observed data by detecting the changes in the patterns contained therein and brought about a higher average compression ratio, 1.6% more than WT (level 5). DAPs provides the advantages of (a) being capable of automatically detecting meaningful patterns, (b) being parameter semi-free, and (c) resulting in a huge reduction in data storage. These findings provide possible applications in the use of various medical devices that produce vast amounts of physiological data that should be monitored.

Original languageEnglish
Pages (from-to)517-536
Number of pages20
JournalJournal of Information Science and Engineering
Volume33
Issue number2
DOIs
Publication statusPublished - 2017 Mar 1

Fingerprint

time series
Time series
Electroencephalography
Chaos theory
Data mining
Data storage equipment
data storage
chaos
knowledge
assistance
segmentation
time

Keywords

  • Chaos population model
  • Compression
  • Electroencephalography
  • Gray-box model
  • Minimum description length
  • Parameter estimation
  • Segmentation

ASJC Scopus subject areas

  • Software
  • Human-Computer Interaction
  • Hardware and Architecture
  • Library and Information Sciences
  • Computational Theory and Mathematics

Cite this

DAPs : Mining using change-point detection of epileptic activity time series data. / Kim, Sun Hee; Li, Lei; Faloutsos, Christos; Yang, Hyung Jeong; Lee, Seong Whan.

In: Journal of Information Science and Engineering, Vol. 33, No. 2, 01.03.2017, p. 517-536.

Research output: Contribution to journalArticle

Kim, Sun Hee ; Li, Lei ; Faloutsos, Christos ; Yang, Hyung Jeong ; Lee, Seong Whan. / DAPs : Mining using change-point detection of epileptic activity time series data. In: Journal of Information Science and Engineering. 2017 ; Vol. 33, No. 2. pp. 517-536.
@article{bc56dc6a1ff3467ba8350dcf911cc3df,
title = "DAPs: Mining using change-point detection of epileptic activity time series data",
abstract = "The goal of this study is to mine meaningful patterns effectively and efficiently via change-point detection of the time series data, with the assistance of domain knowledge and observed data. With those patterns, our method can do segmentation and compression. We developed a novel gray-box approach for mining such data: Domain Assisted Parameter semi-free wave mining (DAPs). DAPs is intended for mining time series with rich domain-specific knowledge based on a chaos model. Specifically, it automatically detects a change-point of time sequences, respecting the minimal description length principle. And the time sequence is segmented based on the detected change-point, and each segment is fitted with a consistent model. The experimental results using both synthetic and real EEG data indicated that the developed method offers a significant improvement in segmentation and compression via pattern detection over other existing methods. DAPs reduced the number of bits of the observed data by detecting the changes in the patterns contained therein and brought about a higher average compression ratio, 1.6{\%} more than WT (level 5). DAPs provides the advantages of (a) being capable of automatically detecting meaningful patterns, (b) being parameter semi-free, and (c) resulting in a huge reduction in data storage. These findings provide possible applications in the use of various medical devices that produce vast amounts of physiological data that should be monitored.",
keywords = "Chaos population model, Compression, Electroencephalography, Gray-box model, Minimum description length, Parameter estimation, Segmentation",
author = "Kim, {Sun Hee} and Lei Li and Christos Faloutsos and Yang, {Hyung Jeong} and Lee, {Seong Whan}",
year = "2017",
month = "3",
day = "1",
doi = "10.1688/JISE.2017.33.2.14",
language = "English",
volume = "33",
pages = "517--536",
journal = "Journal of Information Science and Engineering",
issn = "1016-2364",
publisher = "Institute of Information Science",
number = "2",

}

TY - JOUR

T1 - DAPs

T2 - Mining using change-point detection of epileptic activity time series data

AU - Kim, Sun Hee

AU - Li, Lei

AU - Faloutsos, Christos

AU - Yang, Hyung Jeong

AU - Lee, Seong Whan

PY - 2017/3/1

Y1 - 2017/3/1

N2 - The goal of this study is to mine meaningful patterns effectively and efficiently via change-point detection of the time series data, with the assistance of domain knowledge and observed data. With those patterns, our method can do segmentation and compression. We developed a novel gray-box approach for mining such data: Domain Assisted Parameter semi-free wave mining (DAPs). DAPs is intended for mining time series with rich domain-specific knowledge based on a chaos model. Specifically, it automatically detects a change-point of time sequences, respecting the minimal description length principle. And the time sequence is segmented based on the detected change-point, and each segment is fitted with a consistent model. The experimental results using both synthetic and real EEG data indicated that the developed method offers a significant improvement in segmentation and compression via pattern detection over other existing methods. DAPs reduced the number of bits of the observed data by detecting the changes in the patterns contained therein and brought about a higher average compression ratio, 1.6% more than WT (level 5). DAPs provides the advantages of (a) being capable of automatically detecting meaningful patterns, (b) being parameter semi-free, and (c) resulting in a huge reduction in data storage. These findings provide possible applications in the use of various medical devices that produce vast amounts of physiological data that should be monitored.

AB - The goal of this study is to mine meaningful patterns effectively and efficiently via change-point detection of the time series data, with the assistance of domain knowledge and observed data. With those patterns, our method can do segmentation and compression. We developed a novel gray-box approach for mining such data: Domain Assisted Parameter semi-free wave mining (DAPs). DAPs is intended for mining time series with rich domain-specific knowledge based on a chaos model. Specifically, it automatically detects a change-point of time sequences, respecting the minimal description length principle. And the time sequence is segmented based on the detected change-point, and each segment is fitted with a consistent model. The experimental results using both synthetic and real EEG data indicated that the developed method offers a significant improvement in segmentation and compression via pattern detection over other existing methods. DAPs reduced the number of bits of the observed data by detecting the changes in the patterns contained therein and brought about a higher average compression ratio, 1.6% more than WT (level 5). DAPs provides the advantages of (a) being capable of automatically detecting meaningful patterns, (b) being parameter semi-free, and (c) resulting in a huge reduction in data storage. These findings provide possible applications in the use of various medical devices that produce vast amounts of physiological data that should be monitored.

KW - Chaos population model

KW - Compression

KW - Electroencephalography

KW - Gray-box model

KW - Minimum description length

KW - Parameter estimation

KW - Segmentation

UR - http://www.scopus.com/inward/record.url?scp=85052761828&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85052761828&partnerID=8YFLogxK

U2 - 10.1688/JISE.2017.33.2.14

DO - 10.1688/JISE.2017.33.2.14

M3 - Article

VL - 33

SP - 517

EP - 536

JO - Journal of Information Science and Engineering

JF - Journal of Information Science and Engineering

SN - 1016-2364

IS - 2

ER -