Table 11 Performance comparison with state-of-the-art methods. \(\dagger\): CCCS-CIC-AndMal-2020, \(\ddagger\): KronoDroid. Best in bold. Comparability note: Direct numerical comparison is methodologically valid only among methods evaluated on the same dataset with identical class configurations, feature types, and evaluation protocols. Methods evaluated on different datasets or with different class granularities (e.g., few-shot methods on custom datasets, quantum methods on IoT/SDN datasets) are included for contextual positioning within the broader literature and should be interpreted as reference points rather than strict performance benchmarks.
Method | Reference | Dataset | Acc. | F1 | Features |
|---|---|---|---|---|---|
Methods on CCCS-CIC-AndMal-2020 | |||||
CNN-LSTM Ensemble\(^\dagger\) | Nazim et al.14 | CIC-2020 | 95.36% | 85.0% | Multimodal |
SynDroid\(^\dagger\) | Li et al.5 | CIC-2020 | 93.2% | – | 489 |
Random Forest\(^\dagger\) | Ababneh et al.2 | CIC-2020 | 99.0% | – | 27 |
Dynamic RF-PCA\(^\dagger\) | Al-Sraratee et al.17 | CIC-2020 | 98.5% | 97.0% | 33 |
XGB-AGA\(^\dagger\) | Hammood et al.18 | CIC-2020 | 99.82% | 99.82% | 4,699 |
FSSDroid\(^\dagger\) | Polatidis et al.19 | CIC-2020 | 99.0% | – | 9–27 |
Game APK Detection\(^\dagger\) | Sanamontre et al.20 | CIC-2020 | 94.78% | 94.83% | 9,503 |
HQCNN (12-class)\(^\dagger\) | Sridevi et al.28 | CIC-2020 | 95.13% | – | Wavelet |
Methods on KronoDroid | |||||
Decision Tree\(^\ddagger\) | Guerra-Manzanares10 | KronoDroid | 99.0% | – | 489 |
Random Forest\(^\ddagger\) | Guerra-Manzanares10 | KronoDroid | 99.0% | – | 489 |
FSSDroid\(^\ddagger\) | Polatidis et al.19 | KronoDroid | 99.0% | – | 19–28 |
Obf-Droid\(^\ddagger\) | Aurangzeb et al.21 | KronoDroid | 95.0% | 95.0% | 20 |
DW-FedAvg | Wajahat et al.22 | Malgenome | 99.69% | 99.5% | All |
Few-Shot and Meta-Learning Methods | |||||
SIMPLE | Wang et al.7 | APIMDS | 90.0% | – | API seq. |
Siamese Network | Bai et al.3 | Custom | 87.5% | – | – |
Meta-MAMC | Li et al.8 | Drebin | 97.8% | 97.2% | Static |
FC-Net | Xu et al.26 | Network | 99.62% | – | Traffic |
Quantum and Deep Learning Methods | |||||
QSVM/QCNN | Kalinin et al.12 | IoT-NID | 98.0% | – | 58 |
GDroid (GCN) | Gao et al.1 | Drebin | 98.99% | 97.0% | Graph |
HQCNN (Binary) | Sridevi et al.28 | SDN-DDoS | 99.86% | 99.88% | Wavelet |
Proposed Method | |||||
Proto-QE\(^\dagger\) | This work | CIC-2020 | 99.70% | 99.86% | 51 |
Proto-QE\(^\ddagger\) | This work | KronoDroid | 99.33% | 99.25% | 29 |
