A Novel Quick-Response Eigenface Analysis Scheme for Brain-Computer Interfaces

Abstract

The brain-computer interface (BCI) is used to understand brain activities and external bodies with the help of the motor imagery (MI). As of today, the classification results for EEG 4 class BCI competition dataset have been improved to provide better classification accuracy of the brain computer interface systems (BCIs). Based on this observation, a novel quick-response eigenface analysis (QR-EFA) scheme for motor imagery is proposed to improve the classification accuracy for BCIs. Thus, we considered BCI signals in standardized and sharable quick response (QR) image domain; then, we systematically combined EFA and a convolution neural network (CNN) to classify the neuro images. To overcome a non-stationary BCI dataset available and non-ergodic characteristics, we utilized an effective neuro data augmentation in the training phase. For the ultimate improvements in classification performance, QR-EFA maximizes the similarities existing in the domain-, trial-, and subject-wise directions. To validate and verify the proposed scheme, we performed an experiment on the BCI dataset. Specifically, the scheme is intended to provide a higher classification output in classification accuracy performance for the BCI competition 4 dataset 2a (C4D2a_4C) and BCI competition 3 dataset 3a (C3D3a_4C). The experimental results confirm that the newly proposed QR-EFA method outperforms the previous the published results, specifically from 85.4% to 97.87% ± 0.75 for C4D2a_4C and 88.21% ± 6.02 for C3D3a_4C. Therefore, the proposed QR-EFA could be a highly reliable and constructive framework for one of the MI classification solutions for BCI applications.

Keywords: eigenface analysis; image data augmentation; motor imagery classification; quick response neuro images; standardized and sharable quick response eigenfaces.

Figure 1

The BCI layer model compared to OSI network model.

Figure 2

Overall collaboration between QR-EFA and EFA.

Figure 3

The EFA algorithm procedure.

Figure 4

Data analysis on the viewpoint direction.

Figure 5

Overall flowchart for QR-EFA (solid line) including EFA (dashed line). Cross reference in the above data flowchart for step 1~11.

Figure 6

The CNN for neuro images in QR-EFA.

Figure 7

An original eigenface images of subjects for training data for 4 classes: left, right, feet, and tongue after whitening (12 × 10: = 30 × 4) in C3D3a_4C.

Figure 8

Some selected sample QR training images of data augmentation for the class of (a) left and (b) tongue (12 × 10) in C3D3a_4C.

Figure 9

Primitive raw images for EEG data signal for the selected first and last trials (80 × 60) in C3D3a_4C.

Figure 10

Training QR code realization of subject 1 at trial 1 for 4 classes: left, right, feet, and tongue (12 × 10) in C3D3a_4C.

Figure 11

Data graph of the CNN cost function evolution vs. number of epochs in C3D3a_4C.

Figure 12

An original eigenface images of subjects for training data for 4 classes: left, right, feet, and tongue after whitening (18 × 16: = 72 × 4) in C4D2a_4C.

Figure 13

Some selected sample QR training images of data augmentation for the classes of (a) left and (b) tongue (18 × 16 in C4D2a_4C).

Figure 14

Primitive raw images for EEG data signal for the selected first and last trials (50 × 50) in C4D2a_4C.

Figure 15

Training QR code realization of subject 1 at trial 1 for 4 classes: left, right, feet, and tongue (18 × 16) in C4D2a_4C.

Figure 16

Data graph of the CNN cost function evolution vs. number of epochs in C4D2a_4C.