Table 4 Hyper parameter configuration of models.
From: Robust IoT security using isolation forest and one class SVM algorithms
Model | Tuning method | Hyper parameters considered | Best parameters selected |
|---|---|---|---|
One-Class SVM | Grid Search (threefold CV on normal subset) | kernel: [‘rbf’] gamma: [‘scale’] nu: [0.1, 0.5] | kernel = ‘rbf’, gamma = ‘scale’, nu = 0.1 |
Isolation Forest | Manual | n_estimators: [100] contamination: ‘auto’ | n_estimators = 100, contamination = ‘auto’ |
