Abstract

A tree-ensemble method, referred to as time series forest (TSF), is proposed for time series classification. TSF employs a combination of entropy gain and a distance measure, referred to as the Entrance (entropy and distance) gain, for evaluating the splits. Experimental studies show that the Entrance gain improves the accuracy of TSF. TSF randomly samples features at each tree node and has computational complexity linear in the length of time series, and can be built using parallel computing techniques. The temporal importance curve is proposed to capture the temporal characteristics useful for classification. Experimental studies show that TSF using simple features such as mean, standard deviation and slope is computationally efficient and outperforms strong competitors such as one-nearest-neighbor classifiers with dynamic time warping.

Original languageEnglish (US)
Pages (from-to)142-153
Number of pages12
JournalInformation Sciences
Volume239
DOIs
StatePublished - Aug 1 2013

Fingerprint

Feature Extraction
Feature extraction
Time series
Entropy
Experimental Study
Dynamic Time Warping
Ensemble Methods
Distance Measure
Parallel processing systems
Parallel Computing
Standard deviation
Computational complexity
Nearest Neighbor
Slope
Computational Complexity
Classifiers
Classifier
Curve
Vertex of a graph

Keywords

  • Decision tree
  • Ensemble
  • Entrance gain
  • Interpretability
  • Large margin
  • Time series classification

ASJC Scopus subject areas

  • Artificial Intelligence
  • Software
  • Control and Systems Engineering
  • Theoretical Computer Science
  • Computer Science Applications
  • Information Systems and Management

Cite this

A time series forest for classification and feature extraction. / Deng, Houtao; Runger, George; Tuv, Eugene; Vladimir, Martyanov.

In: Information Sciences, Vol. 239, 01.08.2013, p. 142-153.

Research output: Contribution to journalArticle

Deng, Houtao ; Runger, George ; Tuv, Eugene ; Vladimir, Martyanov. / A time series forest for classification and feature extraction. In: Information Sciences. 2013 ; Vol. 239. pp. 142-153.
@article{0080086ae46644feb736c2d46a6b1846,
title = "A time series forest for classification and feature extraction",
abstract = "A tree-ensemble method, referred to as time series forest (TSF), is proposed for time series classification. TSF employs a combination of entropy gain and a distance measure, referred to as the Entrance (entropy and distance) gain, for evaluating the splits. Experimental studies show that the Entrance gain improves the accuracy of TSF. TSF randomly samples features at each tree node and has computational complexity linear in the length of time series, and can be built using parallel computing techniques. The temporal importance curve is proposed to capture the temporal characteristics useful for classification. Experimental studies show that TSF using simple features such as mean, standard deviation and slope is computationally efficient and outperforms strong competitors such as one-nearest-neighbor classifiers with dynamic time warping.",
keywords = "Decision tree, Ensemble, Entrance gain, Interpretability, Large margin, Time series classification",
author = "Houtao Deng and George Runger and Eugene Tuv and Martyanov Vladimir",
year = "2013",
month = "8",
day = "1",
doi = "10.1016/j.ins.2013.02.030",
language = "English (US)",
volume = "239",
pages = "142--153",
journal = "Information Sciences",
issn = "0020-0255",
publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - A time series forest for classification and feature extraction

AU - Deng, Houtao

AU - Runger, George

AU - Tuv, Eugene

AU - Vladimir, Martyanov

PY - 2013/8/1

Y1 - 2013/8/1

N2 - A tree-ensemble method, referred to as time series forest (TSF), is proposed for time series classification. TSF employs a combination of entropy gain and a distance measure, referred to as the Entrance (entropy and distance) gain, for evaluating the splits. Experimental studies show that the Entrance gain improves the accuracy of TSF. TSF randomly samples features at each tree node and has computational complexity linear in the length of time series, and can be built using parallel computing techniques. The temporal importance curve is proposed to capture the temporal characteristics useful for classification. Experimental studies show that TSF using simple features such as mean, standard deviation and slope is computationally efficient and outperforms strong competitors such as one-nearest-neighbor classifiers with dynamic time warping.

AB - A tree-ensemble method, referred to as time series forest (TSF), is proposed for time series classification. TSF employs a combination of entropy gain and a distance measure, referred to as the Entrance (entropy and distance) gain, for evaluating the splits. Experimental studies show that the Entrance gain improves the accuracy of TSF. TSF randomly samples features at each tree node and has computational complexity linear in the length of time series, and can be built using parallel computing techniques. The temporal importance curve is proposed to capture the temporal characteristics useful for classification. Experimental studies show that TSF using simple features such as mean, standard deviation and slope is computationally efficient and outperforms strong competitors such as one-nearest-neighbor classifiers with dynamic time warping.

KW - Decision tree

KW - Ensemble

KW - Entrance gain

KW - Interpretability

KW - Large margin

KW - Time series classification

UR - http://www.scopus.com/inward/record.url?scp=84876948280&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84876948280&partnerID=8YFLogxK

U2 - 10.1016/j.ins.2013.02.030

DO - 10.1016/j.ins.2013.02.030

M3 - Article

AN - SCOPUS:84876948280

VL - 239

SP - 142

EP - 153

JO - Information Sciences

JF - Information Sciences

SN - 0020-0255

ER -