Table 3 Comparison with the Baseline and State-of-the-Art Methods for EEG200. * means the method is reproduced and reported by CAW-MASA-STST6, and # means the method is reproduced by LMDA36 and is tested in our environment.
From: A hybrid local-global neural network for visual classification using raw EEG signals
Method | Accuracy (%) | |
|---|---|---|
2-class | 10-class | |
Linear | 64.84 ± 0.37 | 24.28 ± 0.31 |
CNN | 60.50 ± 0.26 | 18.95 ± 0.21 |
LSTM | 64.33 ± 0.35 | 23.15 ± 0.12 |
Transformer | 61.70 ± 0.28 | 22.89 ± 0.22 |
EEGNet# | 63.30 ± 0.40 | 22.63 ± 0.25 |
ShallowConvNet# | 61.01 ± 0.45 | 20.61 ± 0.23 |
LDA13 | 56.14 | 14.40 |
LSTM-CNN* | 64.85 | 24.76 |
ShallowConvNet* | 65.53 | 25.13 |
CAW-MASA-STST6 | 68.33 | 31.37 |
Ours | 69.08 ± 0.45 | 29.39 ± 0.19 |
