Affordable Access

Dynamic models for nonstationary signal segmentation.

Authors
  • Penny, W D
  • Roberts, S J
Type
Published Article
Journal
Computers and biomedical research, an international journal
Publication Date
Dec 01, 1999
Volume
32
Issue
6
Pages
483–502
Identifiers
PMID: 10587467
Source
Medline
License
Unknown

Abstract

This paper investigates Hidden Markov Models (HMMs) in which the observations are generated from an autoregressive (AR) model. The overall model performs nonstationary spectral analysis and automatically segments a time series into discrete dynamic regimes. Because learning in HMMs is sensitive to initial conditions, we initialize the HMM model with parameters derived from a cluster analysis of Kalman filter coefficients. An important aspect of the Kalman filter implementation is that the state noise is estimated on-line. This allows for an initial estimation of AR parameters for each of the different dynamic regimes. These estimates are then fine-tuned with the HMM model. The method is demonstrated on a number of synthetic problems and on electroencephalogram data.

Report this publication

Statistics

Seen <100 times