Table 5. SVM with different kernels using 6 features.
Different SVM kernels | Training accuracy (%) | Testing accuracy(%) |
---|---|---|
SVM (kernel = linear, c = 1) | 86.30 | 86.96 |
SVM (kernel = RBF, c = 0.9, γ = 0.1) | 87.55 | 82.60 |
SVM (kernel = poly, c = 1, degree = 8) | 71.78 | 71.01 |
SVM (kernel = sigmoid, c = 1) | 81.74 | 82.60 |