Table 2 Performance for each classifier in all classification problems with and without feature selection
From: Applying nonlinear measures to the brain rhythms: an effective method for epilepsy diagnosis
Classifiers | Sub-Sets | |||||
|---|---|---|---|---|---|---|
Linear SVM | RBF SVM | MLPNN | ||||
Performance with Feature Selection (No. of features) | Performance without Feature Selection (%) | Performance with Feature Selection (No. of features) | Performance without Feature Selection (%) | Performance with Feature Selection (No. of features) | Performance without Feature Selection (%) | |
ABCD/E | 98.21% (27) | 98.09 | 99.2% (23) | 99.05 | 99.91% (22) | 99.69 |
AB/CD/E | 94.54% (25) | 94.41 | 96.65% (20) | 96.02 | 98.19% (26) | 96.86 |
A/D/E | 94.98% (19) | 94.85 | 96.67% (18) | 95.96 | 98.5% (27) | 98.31 |
A/E | 100% (6) | 99.97 | 100% (5) | 100 | 100% (5) | 99.5 |
D/E | 97.56% (26) | 97.34 | 98.41% (19) | 98.25 | 99.84% (23) | 99.66 |