Ransomware detection on Custom threat-informed dynamic behavioral dataset
99.1AccuracyTL-RL-FusionNet
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| TL-RL-FusionNetAnalysis Technique=Dynamic, Feature Extractor(s)=EfficientNetB0 + InceptionV3, RL-based Sample Weighting=Yes, Classifier=RL enhanced Residual MLP, Training Time (s)=13.90, Prediction Time (s)=0.23, Memory Usage (MB)=0.662026.04 | 99.1 | 70 | 20 | |
| Bi-KNNAnalysis Technique=Static, Feature Extractor(s)=VGG-16 + ResNet50, RL-based Sample Weighting=No, Classifier=Bi-KNN, Training Time (s)=18.9282026.04 | 99 | — | — | |
| XGBoostAnalysis Technique=Static, Feature Extractor(s)=DenseNet-121, RL-based Sample Weighting=No, Classifier=XGBoost2026.04 | 98.58 | — | — | |
| MLPAnalysis Technique=Static, Feature Extractor(s)=VGG16, VGG19, ResNet50, InceptionV3, RL-based Sample Weighting=No, Classifier=MLP, Prediction Time (s)=0.4712026.04 | 98.55 | — | — | |
| CyberForce (FL and RL)Analysis Technique=Dynamic, Classifier=FL and RL2026.04 | 98 | — | — | |
| SVMAnalysis Technique=Static, Feature Extractor(s)=VGG-16, RL-based Sample Weighting=No, Classifier=SVM2026.04 | 92.97 | — | — | |
| DRL (A2C)Analysis Technique=Static, RL-based Sample Weighting=No, Classifier=DRL (A2C, DQN, DDQN, PPO), Training Time (s)=168.27, Prediction Time (s)=0.762026.04 | 89.78 | — | — |