Bilinear common spatial pattern for single-trial ERP-based rapid serial visual presentation triage

  • Yu K
  • Shen K
  • Shao S
 et al. 
  • 21

    Readers

    Mendeley users who have this article in their library.
  • 14

    Citations

    Citations of this article.

Abstract

Common spatial pattern (CSP) analysis is a useful tool for the feature extraction of event-related potentials (ERP). However, CSP is essentially time invariant, and thus unable to exploit the temporal information of ERP. This paper proposes a variant of CSP, namely bilinear common spatial pattern (BCSP), which is capable of accommodating both spatial and temporal information. BCSP generalizes CSP through iteratively optimizing bilinear filters. These bilinear filters constitute a spatio-temporal subspace in which the separation between two conditions is maximized. The method is unique in the sense that it is mathematically intuitive and simple, as all the bilinear filters are obtained by maximizing the power ratio as CSP does. The proposed method was evaluated on 20 subjects' ERP data collected in rapid serial visual presentation triage experiments. The results show that BCSP achieved significantly higher average test accuracy (12.3% higher, p < 0.001).

Get free article suggestions today

Mendeley saves you time finding and organizing research

Sign up here
Already have an account ?Sign in

Find this document

Get full text

Authors

  • K. Yu

  • K. Shen

  • S. Shao

  • W. C. Ng

  • X. Li

Cite this document

Choose a citation style from the tabs below

Save time finding and organizing research with Mendeley

Sign up for free