Intrusion Detection on NSL-KDD (test)
99.2RecallStacking-LR
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Stacking-LRBase Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=5, Stacking=3-fold cross-validation2026.05 | 99.2 | — | — | 0.9 | 99.2 | — | — | |
| Stacking-MLPBase Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=5, Stacking=3-fold cross-validation2026.05 | 99.2 | — | — | 0.8 | 99.2 | — | — | |
| VCG (Ours)Base Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=52026.05 | 99.1 | — | — | 0.9 | 99.2 | — | — | |
| BrierBase Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=52026.05 | 99.1 | — | — | 1 | 99.2 | — | — | |
| Best IndividualBase Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=52026.05 | 99.1 | — | — | 0.9 | 99.2 | — | — | |
| ExternalityBase Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=52026.05 | 99 | — | — | 1 | 99.2 | — | — | |
| Conf-WeightedBase Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=52026.05 | 99 | — | — | 1 | 99.2 | — | — | |
| Log-OddsBase Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=52026.05 | 99 | — | — | 1 | 99.2 | — | — | |
| Majority VoteBase Learners=LightGBM/CatBoost/XGBoost/RF/MLP, Feature Partition Size=60%, Seeds=10, Agents=52026.05 | 98.9 | — | — | 1.1 | 99.1 | — | — | |
| MixGANTrain Ratio=90%2025.08 | 96.4 | 95.7 | 94.4 | — | 95.2 | — | 94.4 | |
| MixGANTrain Ratio=80%2025.08 | 95.3 | 94.3 | 92.7 | — | 93.8 | — | 92.7 | |
| MixGANTrain Ratio=70%2025.08 | 92.6 | 93.1 | 92 | — | 92.3 | — | 92 | |
| MixGANTrain Ratio=60%2025.08 | 91.4 | 92.3 | 91.4 | — | 91.4 | — | 91.4 | |
| FACVO-DNFNTrain Ratio=90%2025.08 | 90.9 | 93 | 87.5 | — | 89.1 | — | 92.9 | |
| GHLBO+DSATrain Ratio=90%2025.08 | 90.9 | 91.4 | 90.2 | — | 90.2 | — | 90.1 | |
| GHLBO+DSATrain Ratio=80%2025.08 | 90.7 | 90.8 | 88.3 | — | 89.5 | — | 88 | |
| GHLBO+DSATrain Ratio=60%2025.08 | 89.5 | 89.9 | 86.2 | — | 87.8 | — | 85.7 | |
| GHLBO+DSATrain Ratio=70%2025.08 | 89.5 | 90.2 | 87.1 | — | 88.3 | — | 86.7 | |
| FACVO-DNFNTrain Ratio=80%2025.08 | 88.3 | 92.1 | 86.6 | — | 87.4 | — | 90 | |
| FT-DBNTrain Ratio=90%2025.08 | 87.4 | 90.9 | 85.5 | — | 86.4 | — | 88.9 | |
| HLBO+DSATrain Ratio=90%2025.08 | 86.8 | 89.7 | 87.9 | — | 87.4 | — | 88.1 | |
| HLBO+DSATrain Ratio=80%2025.08 | 86.5 | 89.1 | 88.2 | — | 87.3 | — | 88.4 | |
| FACVO-DNFNTrain Ratio=70%2025.08 | 86.5 | 91.1 | 85.6 | — | 86.1 | — | 89.7 | |
| FT-DBNTrain Ratio=80%2025.08 | 86.3 | 90.2 | 84.8 | — | 85.5 | — | 87.8 | |
| HLBO+DSATrain Ratio=70%2025.08 | 86.2 | 84.3 | 86.5 | — | 86 | — | 87.2 | |
| HLBO+DSATrain Ratio=60%2025.08 | 85.6 | 85.7 | 87.7 | — | 86.1 | — | 87.9 | |
| FACVO-DNFNTrain Ratio=60%2025.08 | 84.9 | 90.4 | 85 | — | 84.9 | — | 89.3 | |
| LUCIDTrain Ratio=90%2025.08 | 84.2 | 84.7 | 82.6 | — | 83.4 | — | 82.3 | |
| FT-DBNTrain Ratio=70%2025.08 | 84 | 89 | 83.7 | — | 83.9 | — | 86.8 | |
| LUCIDTrain Ratio=80%2025.08 | 83.2 | 83.5 | 81.5 | — | 82.3 | — | 81.1 | |
| LUCIDTrain Ratio=70%2025.08 | 82.6 | 81.3 | 80.1 | — | 81.3 | — | 79.5 | |
| FT-DBNTrain Ratio=60%2025.08 | 82.6 | 88.4 | 83.1 | — | 82.9 | — | 85.7 | |
| LUCIDTrain Ratio=60%2025.08 | 81.9 | 80.6 | 78.1 | — | 80 | — | 77.1 | |
| FedEPDistribution=non-IID, Client Nodes=202026.02 | 81.75 | 84.24 | 89.66 | 18.25 | 85.52 | — | — | |
| FedPGDistribution=non-IID, Client Nodes=202026.02 | 81.3 | 84.09 | 89.78 | 18.7 | 85.33 | — | — | |
| CMDNumber of features=40, Classifier=3-hidden-layer MLP2026.06 | — | — | — | — | — | 31.24 | — | |
| Fisher ScoreNumber of features=40, Classifier=3-hidden-layer MLP2026.06 | — | — | — | — | — | 32.09 | — | |
| Mutual InformationNumber of features=40, Classifier=3-hidden-layer MLP2026.06 | — | — | — | — | — | 32.74 | — | |
| nCMDNumber of features=40, Classifier=3-hidden-layer MLP2026.06 | — | — | — | — | — | 34.29 | — | |
| Pearson CorrelationNumber of features=40, Classifier=3-hidden-layer MLP2026.06 | — | — | — | — | — | 28.81 | — | |
| Variance ThresholdNumber of features=40, Classifier=3-hidden-layer MLP2026.06 | — | — | — | — | — | 28.35 | — |