Table 1.
Corrected Friedman post-hoc p-values
|
feature
architecture sample ratio error margin |
OV
1 1% 10% |
OV
1 1% 5% |
OV
1 10% 1% |
OV
1 5% 1% |
OV+A
2 1% 10% |
OV+A
2 1% 5% |
OV+A
2 10% 5% |
OV+A
2 5% 5% |
|---|---|---|---|---|---|---|---|---|
| OV-1-1%−1% | 0.33 | 1.00 | 1.00 | 1.00 | 0.76 | 0.00 | 0.00 | 0.00 |
| OV-1-1%−10% | 1.00 | 0.63 | 1.00 | 0.00 | 0.00 | 0.00 | 0.00 | |
| OV-1-1%−5% | 1.00 | 1.00 | 0.14 | 0.00 | 0.00 | 0.00 | ||
| OV-1-10%−1% | 1.00 | 0.43 | 0.00 | 0.00 | 0.00 | |||
| OV-1-5%−1% | 0.10 | 0.00 | 0.00 | 0.00 | ||||
| OV+A-2-1%−10% | 0.05 | 0.33 | 0.33 | |||||
| OV+A-2-1%−5% | 1.00 | 1.00 | ||||||
| OV+A-2-10%−5% | 1.00 |
*
insignificant p-values (>0.05) with Bergmann and Hommel correction are highlighted in Bold for pairwise similarities. The top-performing encoder is highlighted in the first row of the matrix.