Malicious URL Detection on Balanced URL dataset (test)
0.964AccuracyProposed Hybrid Model
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Proposed Hybrid ModelArchitecture=Lightweight feedforward neural network, Preprocessing=SMOTE + Isolation Forest, Integration=Tkinter-based GUI2025.11 | 0.964 | 0.957 | 0.952 | 0.954 | 0.973 | |
| CNN (Char-level model)Architecture=Char-level model, Preprocessing=SMOTE + Isolation Forest2025.11 | 0.948 | 0.942 | 0.937 | 0.939 | 0.951 | |
| Random ForestVectorization=HashingVectorizer (n-gram 2,5; features=1000), Preprocessing=SMOTE + Isolation Forest2025.11 | 0.936 | 0.928 | 0.922 | 0.925 | 0.944 | |
| SVM (RBF kernel)Kernel=RBF, Vectorization=HashingVectorizer (n-gram 2,5; features=1000), Preprocessing=SMOTE + Isolation Forest2025.11 | 0.924 | 0.916 | 0.908 | 0.912 | 0.935 | |
| Logistic RegressionVectorization=HashingVectorizer (n-gram 2,5; features=1000), Preprocessing=SMOTE + Isolation Forest2025.11 | 0.913 | 0.901 | 0.896 | 0.898 | 0.921 |