Mining the independent source of ERP components with ICA decomposition

Authors:
Jia-Cai Zhang;Xiao-Jie Zhao;Yi-Jun Liu;Li Yao
Affiliations:
College of Information Science and Technology, Beijing Normal University, Beijing, China;College of Information Science and Technology, Beijing Normal University, Beijing, China;McKnight Brain Institute, University of Florida, Gainesville, FL;College of Information Science and Technology, Beijing Normal University, Beijing, China
Venue:
ISNN'06 Proceedings of the Third international conference on Advances in Neural Networks - Volume Part III
Year:
2006

Citing 4
Cited 0

Independent component analysis using an extended infomax algorithm for mixed subgaussian and supergaussian sources

Neural Computation
Independent component analysis: algorithms and applications

Neural Networks
Recognition and reproduction of gestures using a probabilistic framework combining PCA, ICA and HMM

ICML '05 Proceedings of the 22nd international conference on Machine learning
One-Bit-Matching Conjecture for Independent Component Analysis

Neural Computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

Independent component analysis (ICA) can blindly separates the input ERP data into a sum of temporally independent and spatially fixed components arising from distinct or overlapping brain regions. In this study, we use ICA to illustrate that the P300 components in two ERPs recorded under various conditions or tasks are both mainly contributed from a few independent sources. ICA decomposition also indicates a new method to compare P300 components between two ERPs induced by two related tasks. Our comparisons are made on those independent sources contributed to the P300 components, rather than on the ERP waveforms directly. This novel approach identifies not only the similar or common independent components in both conditions that bring about a common part in ERP time courses, but also those different components induced by the different parts in ERP waveforms. Our study suggests that the ICA method is a useful tool to study the brain dynamics.