Table 4. The classification accuracies of each piece after applied to OVA in the classification of multi-class EEG signals using the Random Forests classifier with and without ADASYN using 5-FCV (the third and fourth approaches).
The obtained piece after OVO | The obtained Classification accuracy (%) Without ADASYN |
The obtained Classification accuracy (%) With ADASYN |
---|---|---|
Class 1-others (the combination of class 2, class 3, class 4, and class 5) | 98.10 | 98.40 |
Class 2-others (the combination of class 1, class 3, class 4, and class 5) | 83.00 | 97.70 |
Class 3-others (the combination of class 1, class 2, class 4, and class 5) | 84.60 | 88.50 |
Class 4-others (the combination of class 1, class 2, class 3, and class 5) | 91.20 | 92.70 |
Class 5-others (the combination of class 1, class 2, class 3, and class 4) | 88.10 | 91.50 |
The overall accuracy (average score) | 89.00 | 91.72 |