Table 3 Performance comparison of different models for motor imagery classification.
Model | Accuracy (%) | Precision (%) | Recall (%) | F1-Score (%) | AUC | Training time (s) |
|---|---|---|---|---|---|---|
Random forest | 92.52 ± 1.23 | 91.87 ± 1.45 | 92.52 ± 1.23 | 92.19 ± 1.34 | 0.982 ± 0.005 | 45.3 ± 2.7 |
SVM | 90.18 ± 1.56 | 89.95 ± 1.78 | 90.18 ± 1.56 | 90.06 ± 1.67 | 0.976 ± 0.007 | 38.7 ± 1.9 |
LSTM | 88.89 ± 2.01 | 88.23 ± 2.34 | 88.89 ± 2.01 | 88.56 ± 2.17 | 0.971 ± 0.009 | 256.4 ± 12.3 |
BiLSTM | 91.75 ± 1.67 | 91.42 ± 1.89 | 91.75 ± 1.67 | 91.58 ± 1.78 | 0.980 ± 0.006 | 342.8 ± 15.6 |
CNN-LSTM | 93.58 ± 1.12 | 93.21 ± 1.35 | 93.58 ± 1.12 | 93.39 ± 1.23 | 0.987 ± 0.004 | 478.2 ± 18.9 |
CNN-LSTM-Attention | 97.25 ± 0.78 | 97.18 ± 0.89 | 97.25 ± 0.78 | 97.21 ± 0.83 | 0.995 ± 0.002 | 523.6 ± 21.5 |
