Skip to main content

Table 1 Offline classification accuracies for the baseline method and the PCA-normalization

From: Principal component based covariate shift adaption to reduce non-stationarity in a MEG-based brain-computer interface

Subject

CV

S1S2

 

Baseline(%)

PCA(%)

Baseline(%)

PCAnorm(%)

A01

88.5 %

90.9 %

86.3 %

88.2 %

A02

86.7 %

90.6 %

82.4 %

94.1 %

A03

92.4 %

93.6 %

77.5 %

96.1 %

A04

92.9 %

94.1 %

90.2 %

90.2 %

A05

94.8 %

97.4 %

80.4 %

95.1 %

A06

89.9 %

90.1 %

63.7 %

88.2 %

A07

96.1 %

97.8 %

91.4 %

98.0 %

A08

99.3 %

98.6 %

91.4 %

98.0 %

A09

69.0 %

81.4 %

69.6 %

74.5 %

A10

98.6 %

98.2 %

96.1 %

99.0 %

mean

90.8 %

93.3 %

83.4 %

92.2 %

  1. Results on the left side(CV) are estimated by a 5x10-fold cross-validation. For results on the right(S1S2), the classifier was trained on session 1 and tested on session 2. Best results are printed bold.