Table 3 Experimental results for combined sessions- Seq (S1S + S2S) based on different pre-trained models
From: An evaluation of transfer learning models in EEG-based authentication
Model | Comparison of classification accuracy (Averaged% ± Standard deviation) | ||||
|---|---|---|---|---|---|
Acc. | Pre. | Sens. | Spec. | F1 | |
GoogLeNet | 97.61 ± 0.34 | 97.93 ± 0.85 | 97.54 ± 0.34 | 99.91 ± 0.001 | 97.45 ± 0.23 |
Inception-V3 | 97.75 ± 0.46 | 97.80 ± 0.54 | 97.69 ± 0.46 | 99.92 ± 0.02 | 97.64 ± 0.47 |
ResNet-50 | 97.37 ± 0.08 | 97.30 ± 0.08 | 97.30 ± 0.08 | 99.91 ± 0.003 | 97.30 ± 0.08 |
ResNet-101 | 97.05 ± 0.23 | 96.97 ± 0.23 | 96.96 ± 0.23 | 99.89 ± 0.01 | 96.96 ± 0.23 |
EfficientNet-B0 | 97.10 ± 0.15 | 97.04 ± 0.15 | 97.02 ± 0.14 | 99.90 ± 0.01 | 97.01 ± 0.13 |
DenseNet-201 | 97.24 ± 0.27 | 97.16 ± 0.27 | 97.16 ± 0.27 | 99.90 ± 0.01 | 97.16 ± 0.27 |