Detecting dynamical change in nonlinear time series

L. M. Hively; P. C. Gailey; V. A. Protopopescu

doi:10.1016/S0375-9601(99)00342-4

Detecting dynamical change in nonlinear time series

L. M. Hively, P. C. Gailey, V. A. Protopopescu

Computational Sciences & Engineering Div

Research output: Contribution to journal › Article › peer-review

44 Scopus citations

Abstract

We present a robust, model-independent technique for measuring changes in the dynamics underlying nonlinear time-serial data. After constructing discrete density distributions of phase-space points on the attractor for time-windowed data sets, we measure the dissimilarity between density distributions via L₁-distance and _χ² statistics. The discriminating power of the new measures is first tested on the Lorenz model and then applied to EEG data to detect the transition between non-seizure and epileptic activity. We find a clear superiority of the new measures in comparison to traditional nonlinear measures as discriminators of changing dynamics.

Original language	English
Pages (from-to)	103-114
Number of pages	12
Journal	Physics Letters, Section A: General, Atomic and Solid State Physics
Volume	258
Issue number	2-3
DOIs	https://doi.org/10.1016/S0375-9601(99)00342-4
State	Published - Jul 19 1999

Funding

This work was sponsored by the U.S. Department of Energy, Office of Energy Management. V.P. was partially supported by the U.S. Department of Energy, Office of Basic Energy Sciences. Oak Ridge National Laboratory is managed for the United States Department of Energy by Lockheed Martin Energy Research Corporation, under Contract No. DE-AC05-96OR22464. We are grateful to members of the staff at the Knoxville Neurology Clinic at St. Mary's Medical Center (Knoxville, TN), including Dr. M.L. Eisenstadt, Dr. R.E. Leppanen, W.W. Holland, and H.L. Leach, who provided expert assistance with the EEG data acquisition and interpretation. We thank N.E. Clapp, Jr. (ORNL) for converting the data from archival VHS tapes to digital form, and for determining the times of the clinical seizures from the paper records.

Funders	Funder number
Office of Basic Energy Sciences
Office of Energy Management
United States Department of Energy
U.S. Department of Energy
Lockheed Martin Corporation

Access to Document

10.1016/S0375-9601(99)00342-4

Cite this

@article{0c5d7383f3f74f65a383f7e2102b36e6,

title = "Detecting dynamical change in nonlinear time series",

abstract = "We present a robust, model-independent technique for measuring changes in the dynamics underlying nonlinear time-serial data. After constructing discrete density distributions of phase-space points on the attractor for time-windowed data sets, we measure the dissimilarity between density distributions via L1-distance and χ2 statistics. The discriminating power of the new measures is first tested on the Lorenz model and then applied to EEG data to detect the transition between non-seizure and epileptic activity. We find a clear superiority of the new measures in comparison to traditional nonlinear measures as discriminators of changing dynamics.",

author = "Hively, {L. M.} and Gailey, {P. C.} and Protopopescu, {V. A.}",

year = "1999",

month = jul,

day = "19",

doi = "10.1016/S0375-9601(99)00342-4",

language = "English",

volume = "258",

pages = "103--114",

journal = "Physics Letters, Section A: General, Atomic and Solid State Physics",

issn = "0375-9601",

publisher = "Elsevier",

number = "2-3",

}

TY - JOUR

T1 - Detecting dynamical change in nonlinear time series

AU - Hively, L. M.

AU - Gailey, P. C.

AU - Protopopescu, V. A.

PY - 1999/7/19

Y1 - 1999/7/19

N2 - We present a robust, model-independent technique for measuring changes in the dynamics underlying nonlinear time-serial data. After constructing discrete density distributions of phase-space points on the attractor for time-windowed data sets, we measure the dissimilarity between density distributions via L1-distance and χ2 statistics. The discriminating power of the new measures is first tested on the Lorenz model and then applied to EEG data to detect the transition between non-seizure and epileptic activity. We find a clear superiority of the new measures in comparison to traditional nonlinear measures as discriminators of changing dynamics.

AB - We present a robust, model-independent technique for measuring changes in the dynamics underlying nonlinear time-serial data. After constructing discrete density distributions of phase-space points on the attractor for time-windowed data sets, we measure the dissimilarity between density distributions via L1-distance and χ2 statistics. The discriminating power of the new measures is first tested on the Lorenz model and then applied to EEG data to detect the transition between non-seizure and epileptic activity. We find a clear superiority of the new measures in comparison to traditional nonlinear measures as discriminators of changing dynamics.

UR - http://www.scopus.com/inward/record.url?scp=0040219905&partnerID=8YFLogxK

U2 - 10.1016/S0375-9601(99)00342-4

DO - 10.1016/S0375-9601(99)00342-4

M3 - Article

AN - SCOPUS:0040219905

SN - 0375-9601

VL - 258

SP - 103

EP - 114

JO - Physics Letters, Section A: General, Atomic and Solid State Physics

JF - Physics Letters, Section A: General, Atomic and Solid State Physics

IS - 2-3

ER -

Detecting dynamical change in nonlinear time series

Abstract

Funding

Access to Document

Other files and links

Fingerprint

Cite this