Dynamics & Sparsity in Latent Threshold Factor Models: A Study in Multivariate EEG Signal Processing

Jouchi Nakajima, Mike West
Bank of Japan, Duke University

Apr 8 2016

We discuss Bayesian analysis of multivariate time series with dynamic factor models that exploit time-adaptive sparsity in model parametrizations via the latent threshold approach. One central focus is on the transfer responses of multiple interrelated series to underlying, dynamic latent factor processes. Structured priors on model hyper-parameters are key to the efficacy of dynamic latent thresholding, and MCMC-based computation enables model fitting and analysis. A detailed case study of electroencephalographic (EEG) data from experimental psychiatry highlights the use of latent threshold extensions of time-varying vector autoregressive and factor models. This study explores a class of dynamic transfer response factor models, extending prior Bayesian modeling of multiple EEG series and highlighting the practical utility of the latent thresholding concept in multivariate, non-stationary time series analysis.

This work was supported in part by a grant from the U.S. National Science Foundation (DMS-1106516). Any opinions, findings and conclusions or recommendations expressed in this work are those of the authors and do not necessarily reflect the views of the NSF or the Bank of Japan.

Supplementary material include these on-line animations of three figures discussed in the case study in the paper.


Dynamic latent factors, Electroconvulsive therapy (ECT), Electroencephalographic (EEG) time series, Impulse response, Latent threshold dynamic models, Multivariate time series, Sparse time-varying loadings, Time-series decomposition, Time-varying...


PDF icon 2013-04.pdf

BibTeX Citation: 

  author = {J. Nakajima and M. West},
  title = {Dynamics and sparsity in latent threshold factor models: {A} study in multivariate {EEG} signal processing},
  journal = {Under review at: Brazilian Journal of Probability and Statistics},
  year = {2016},
  note = {Revised version: April 2016},
  url = {https://stat.duke.edu/research/papers/2013-04}