Automated granule discovery in continuous data for feature selection

A supervised method to enhance distance-based neural network clustering performance by discovering perfect representative neurons

Distance-based neural network clustering requires the intrinsic assumption that a particular neuron in the network represents a cluster centroid. However, not all these neurons can perfectly represent the training data; these neurons can only represent part of the training samples. This paper proposes an effective training data splitting method (TDSM) to find perfect representative neurons and improve the clustering results in a distance-based neutral network without changing the original network’s internal algorithm or the training data quality. The method allows a network with N neurons to be enlarged to a new network with $$m\times N$$ m × N neurons. These neurons represent m subnetworks, and each subnetwork perfectly represents a part of the training set, where the clustering qualification indicators (the purity, normalized mutual information, and adjusted rand index measures) all equal 1. The results are statistically validated with a t test, and we demonstrate that the TDSM performs better than the original clustering paradigm on some real datasets.

Comparative study of attention-related features on attention monitoring systems with a single EEG channel

The easy-to-use attention monitoring systems usually detect the participant's attentional status via processing electroencephalogram (EEG) data recorded from a single FPz channel. But due to the influence of noises and artifacts, the attention-monitoring performance needs to be further improved to suit different individuals and devices. This paper compared the attention-related features extracted using four state-of-the-art methods including delta/beta1 (D/B1), α + β + δ + θ + R, entropy and optimized complex network (OCN). The classification performance was evaluated using receiver operating characteristic (ROC) curves and area under the ROC curves (AUC) on two EEG data acquisition devices, i.e., a BrainAmp device with high precision and a Sichiray device with low cost, respectively. Considering the varied performance on different individuals and devices, this paper proposed a novel Mutual information-based feature fusion (MIFF) method, selecting the optimal combinations of the attention-related features for classification, to enhance the attention detection performance. The experimental results showed that the proposed MIFF method outperformed the state-of-the-art methods regardless of data length on both devices. Especially, the proposed method with data length of 2.5 s achieved an average AUC of 0.8505 on the low-cost Sichiray device, which is 56.08 % higher than that of D/B1, 27.28 % higher than that of α + β + δ + θ + R, 17.42 % higher than that of entropy, and 15.48 % higher than that of OCN.