Table 8.
Accuracy of the AdaBoost classifier trained on the ENDM-based filtered training set with the noise-correction modality, four different metrics (SuMax, UnMax, SuSum, UnSum) and three different classifiers (Bagging, AdaBoost, and k-NN with ). The values in brackets are the detected noise ratio. The best results are marked in bold.
| SuMax | UnMax | SuSum | UnSum | ||
|---|---|---|---|---|---|
| Letter | Bagging | 50.94 (1%) | 50.85 (1%) | 46.02 (0%) | 50.88(1%) |
| AdaBoost | 50.44 (1%) | 50.44 (1%) | 48.40 (0%) | 50.32 (1%) | |
| k-NN | 43.56 (9%) | 46.11 (7%) | 41.81 (15%) | 44.00 (8%) | |
| Optdigits | Bagging | 89.80 (0%) | 90.10 (0%) | 89.87 (0%) | 89.80 (0%) |
| AdaBoost | 93.23 (16%) | 92.30 (15%) | 92.21 (18%) | 91.33 (17%) | |
| k-NN | 92.36 (17%) | 92.73 (16%) | 92.17 (19%) | 90.51 (22%) | |
| Pendigit | Bagging | 90.89 (6%) | 90.76 (3%) | 90.04 (0%) | 91.05 (6%) |
| AdaBoost | 93.57 (19%) | 91.84 (16%) | 93.64 (18%) | 91.57 (12%) | |
| k-NN | 92.11 (15%) | 91.00 (14%) | 93.08 (18%) | 91.11 (14%) | |
| Statlog | Bagging | 86.48 (14%) | 86.02 (16%) | 84.83 (0%) | 86.50 (12%) |
| AdaBoost | 87.80 (20%) | 86.39 (9%) | 88.15 (19%) | 86.38 (13%) | |
| k-NN | 86.58 (18%) | 86.16 (16%) | 86.92 (23%) | 86.24 (16%) | |
| Vehicle | Bagging | 73.90 (1%) | 73.75 (1%) | 73.50 (0%) | 72.95 (0%) |
| AdaBoost | 73.85 (17%) | 72.00 (20%) | 73.55 (17%) | 72.80 (23%) | |
| k-NN | 73.85 (11%) | 72.90 (2%) | 73.05 (8%) | 72.95 (2%) |