Table 2.
Test accuracy scores for all the classification methods used on fBIRN and COBRE datasets. See Figure 3 and subsection 2.6 for visualization and detailed explanation of these methods respectively.
| Method | Architecture Category | Test Accuracy (fBIRN) (mean ± std) | Test Accuracy (COBRE) (mean ± std) |
|---|---|---|---|
|
| |||
| SVM | Linear | 79.492 ± 4.886 | 73.688 ± 6.646 |
| LOG | Linear | 79.651 ±4.430 | 72.062 ± 8.254 |
| RFC | Linear | 74.032 ± 4.674 | 69.625 ± 8.818 |
|
| |||
| MLP | Neural Net | 78.603 ± 5.813 | 73.312 ±7.132 |
| FNN | Neural Net | 78.667 ±5.189 | 69.062 ± 6.967 |
| BCNN | Neural Net | 75.429 ± 4.701 | 71.812 ±6.011 |
|
| |||
| UNIF0 | Branched | 78.571 ±4.924 | 71.125 ±8.138 |
| UNIF1 | Branched | 77.302 ±4.515 | 69.062 ±6.351 |
| UNIF2 | Branched | 77.270 ± 4.767 | 70.250 ± 6.502 |
|
| |||
| RNDS | Branched + Flexible | 77.048 ±4.910 | 71.250 ±8.077 |
| GRDS | Branched + Flexible | 77.746 ± 4.747 | 72.625 ±6.142 |
| TPE | Branched + Flexible | 81.052 ± 4.515 | 78.188 ± 6.468 |