Research Article
Round Randomized Learning Vector Quantization for Brain Tumor Imaging
Table 4
Average accuracy rates with and without multirandomization for all LVQ sibling and MLP, RBF, SOM, and RF classifiers.
| Classifier | LVQs | MLP | RBF | SOM | RF |
| Datasets | Ave BR | Ave AR | BR | AR | BR | AR | BR | AR | BR | AR | Brain images data (PPUKM) | 79.28 | 85.20 | 56.58 | 67.76 | 59.21 | 71.05 | 70.39 | 75 | 74.34 | 80.92 | Segmented challenge (UCI) | 88.45 | 92.72 | 88.67 | 81.33 | 88.67 | 90.89 | 75.78 | 82.67 | 92 | 95.78 | Segment test (UCI) | 88.27 | 92.39 | 67.49 | 78.19 | 85.19 | 88.07 | 72.02 | 82.3 | 88.48 | 92.18 | Image segmentation (UCI) | 90.33 | 93.15 | 77.2 | 85.71 | 85.14 | 90.48 | 77.34 | 80.66 | 91.92 | 94.66 | Average and STD | 86.58 ± 4.96 | 90.87 ± 3.79 | 72.49 ± 13.69 | 78.25 ± 7.64 | 79.55 ± 13.67 | 85.12 ± 9.46 | 73.88 ± 3.23 | 80.16 ± 3.55 | 86.69 ± 8.39 | 90.89 ± 6.81 |
|
|
(i) BR means before multirandomizing data sets. (ii) AR means after multirandomizing data sets.
|