Table 5.
CNN model performance for various single wavelet transformations applied to input data. Results have been estimated using repeated k-fold validation (5 rounds with 5 folds each).
| Wavelet | Tensor | Overall Acc | Signal Acc | Artefact Acc |
|---|---|---|---|---|
| Number | Depth | ± Std Dev | ± Std Dev | ± Std Dev |
| D2 | 1 | 98.56 ± 0.62 | 98.74 ± 0.80 | 98.41 ± 0.90 |
| D4 | 1 | 98.54 ± 0.48 | 98.51 ± 0.86 | 98.45 ± 0.99 |
| D6 | 1 | 98.48 ± 0.46 | 98.51 ± 0.86 | 98.45 ± 0.99 |
| D8 | 1 | 98.48 ± 0.46 | 98.51 ± 0.86 | 98.45 ± 0.99 |
| D10 | 1 | 98.62 ± 0.49 | 99.01 ± 0.66 | 98.22 ± 0.78 |
| D12 | 1 | 98.63 ± 0.49 | 98.88 ± 0.69 | 98.34 ± 0.73 |
| D14 | 1 | 98.53 ± 0.56 | 98.60 ± 0.91 | 98.45 ± 0.83 |
| D16 | 1 | 98.44 ± 0.56 | 98.60 ± 0.91 | 98.45 ± 0.83 |
| D18 | 1 | 98.38 ± 0.40 | 98.49 ± 0.64 | 98.25 ± 0.94 |
| D20 | 1 | 98.28 ± 0.57 | 98.55 ± 0.83 | 97.99 ± 1.19 |