A Resampling Test for the Total Independence of Stationary Time Series: Application to the Performance Evaluation of ICA Algorithms

Authors:
Juha Karvanen
Affiliations:
Laboratory for Advanced Brain Signal Processing, RIKEN Brain Science Institute, Japan
Venue:
Neural Processing Letters
Year:
2005

Citing 5
Cited 2

Independent component analysis, a new concept?

Signal Processing - Special issue on higher order statistics
An information-maximization approach to blind separation and blind deconvolution

Neural Computation
Blind separation methods based on Pearson system and its extensions

Signal Processing
Characteristic-function-based independent component analysis

Signal Processing - Special section: Security of data hiding technologies
Fast and robust fixed-point algorithms for independent component analysis

IEEE Transactions on Neural Networks

Stability and Chaos of a Class of Learning Algorithms for ICA Neural Networks

Neural Processing Letters
A smoothed bootstrap test for independence based on mutual information

Computational Statistics & Data Analysis

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper addresses the independence testing of stationary time series. We develop a resampling test based on the Kankainen---Ushakov test of total independence. The resampling test, contrary to the original test, can be also applied to the data with a time-structure. The simulation studies demonstrate the good performance of the proposed test even with strongly autocorrelated time series. As an application, we consider biomedical signal processing and independent component analysis (ICA). The independence test can be used as a performance criterion for ICA algorithms. The practical example of performance evaluation deals with the ICA of electroencephalogram (EEG) data.