Table A1.
Accuracy rate results of different classification approaches for flat skin cover.
| Classifiers | Training Accuracy | Validation Accuracy | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Naïve Bayes | 81.72% | 80% | ||||||||
| Neural Network (MLP) | na = 1 | n = 2 | n = 3 | n = 5 | n = 1 | n = 2 | n = 3 | n = 5 | ||
| 33.3% | 30% | 63.3% | 30% | 30% | 25.6% | 53.3% | 26.7% | |||
| SVM with Linear Kernel | Cb = 0.3 | C = 1 | C = 5 | C = 10 | C = 100 | C = 0.3 | C = 1 | C = 5 | C = 10 | C = 100 |
| 88.3% | 90.8% | 93.3% | 94.2% | 97.5% | 78.33% | 80.83% | 84.2% | 81.7% | 85% | |
| SVM with Polynomial Kernel | C = 0.3 | C = 1 | C = 5 | C = 10 | C = 100 | C = 0.3 | C = 1 | C = 5 | C = 10 | C = 100 |
| pc = 2 | 72.5% | 79.2% | 86.7% | 90% | 94.2% | 52.5% | 65% | 77.5% | 77.5% | 81.7% |
| p = 3 | 62.5% | 71.7% | 82.5% | 85% | 92.5% | 41.7% | 46.7% | 67.5% | 68.3% | 81.7% |
| p = 4 | 50% | 61.7% | 72.5% | 78.3% | 88.3% | 37.5% | 47.5% | 50% | 57.5% | 74.2% |
| p = 5 | 43.3% | 53.3% | 69.2% | 71.7% | 85.8% | 34.2% | 42.5% | 46.7% | 46.7% | 61.7% |
| SVM with RBF Kernel | C = 0.3 | C = 1 | C = 5 | C = 10 | C = 100 | C = 0.3 | C = 1 | C = 5 | C = 10 | C = 100 |
| γd = 0.1 | 81.7% | 86.7% | 91.7% | 92.5% | 98.3% | 78.3% | 75% | 84.2% | 83.3% | 88.3% |
| γ = 0.3 | 85.8% | 92.5% | 94.2% | 95.8% | 98.3% | 77.5% | 80.8% | 85% | 85.8% | 90% |
| γ = 0.5 | 90% | 92.5% | 95.8% | 97.5% | 99.2% | 79.2% | 84.2% | 83.3% | 86.7% | 90% |
a
n shows the number of neurons in the hidden layer of a Multi-Layer Perceptron
b
C represents the upper bound parameter of the SVM soft margin
c
p is the degree of the Polynomial kernel
d
γ shows the parameter of SVM Radial Basis Function kernel