Table 4 Comparison of imbalance handling Techniques.
Technique | Description | Advantages | Accuracy | Precision | Recall | F1-Score |
|---|---|---|---|---|---|---|
Class Weighting | Adjusts the loss function to favor minority class. | Simple to implement, integrated into training. | 98.56% | 99.08% | 98.56% | 98.70% |
SMOTE | Generates synthetic samples for minority class. | Improves recall, especially for rare attacks. | 97.83% | 98.30% | 97.84% | 97.87% |
ADASYN | Focuses on generating synthetic data for difficult instances. | Focuses on hard-to-learn instances. | 97.90% | 98.15% | 98.00% | 98.07% |
Focal Loss | Down-weights easy examples, focusing on difficult cases. | Handles imbalance effectively, improves Precision. | 98.12% | 98.36% | 98.05% | 98.10% |
