Classification Approach for Attention Assessment via Singular Spectrum Analysis Based on Single-Channel Electroencephalograms

Abstract

Attention refers to the human psychological ability to focus on doing an activity. The attention assessment plays an important role in diagnosing attention deficit hyperactivity disorder (ADHD). In this paper, the attention assessment is performed via a classification approach. First, the single-channel electroencephalograms (EEGs) are acquired from various participants when they perform various activities. Then, fast Fourier transform (FFT) is applied to the acquired EEGs, and the high-frequency components are discarded for performing denoising. Next, empirical mode decomposition (EMD) is applied to remove the underlying trend of the signals. In order to extract more features, singular spectrum analysis (SSA) is employed to increase the total number of the components. Finally, some typical models such as the random forest-based classifier, the support vector machine (SVM)-based classifier, and the back-propagation (BP) neural network-based classifier are used for performing the classifications. Here, the percentages of the classification accuracies are employed as the attention scores. The computer numerical simulation results show that our proposed method yields a higher classification performance compared to the traditional methods without performing the EMD and SSA.

Keywords: attention assessment; back-propagation neural network; empirical mode decomposition; random forest; single-channel electroencephalograms; singular spectrum analysis; support vector machine.

Figure 1

The diagonal averaging operation.

Figure 2

The block diagram of our proposed method.

Figure 3

The acquisition device.

Figure 4

An example of a sample of the EEG when Danny is conducting the drawing activity under the concentration state.

Figure 5

The denoised signal and the detrended signal.

Figure 6

The obtained brain waves when Danny is conducting the drawing activity under the concentration state.

Figure 7

The four groups of the SSA components.

Figure 8

These two sets of the SSA components decomposed from the brain waves of a detrended EEG.

Figure 9

The procedures for the formation of the random forest.

Figure 10

The BP neural network.

Figure 11

The average percentages of the classification accuracies over both these 12 participants performing the various activities yielded by both our proposed method and the traditional methods with different classifiers under the concentration state.

Figure 12

The average percentages of the classification accuracies over both these 12 participants performing the various activities yielded by both our proposed method and the traditional methods with different classifiers under the immersion state.

Figure 13

The relationship between the percentage of the classification accuracy and the total number of neurons in the hidden layer of the BP neural network in our proposed method trained using the EEGs acquired from Hejian performing the reading activity under the concentration state on the first day.

Figure 14

The relationship between the percentage of the classification accuracy and the total number of neurons in the hidden layer of the BP neural network in our proposed method trained using the EEGs acquired from Hejian performing the reading activity under the concentration state on the second day.

Figure 15

The relationship between the mean squares error and the total number of trees at different total numbers of leaf nodes in the random forest in our proposed method trained using the EEGs acquired from Hejian performing the reading activity under the concentration state on the first day.

Figure 16

The relationship between the mean squares error and the total number of trees at different total numbers of leaf nodes in the random forest in our proposed method trained using the EEGs acquired from Hejian performing the reading activity under the concentration state on the second day.

Figure 17

The percentages of the classification accuracies yielded by our proposed method with three different classifiers trained using the EEGs acquired from Hejian performing the reading activity under the concentration state on both the first day and the second day.