Table 11 Comparison of sample size, classification type, dataset, accuracy, feature extractor, data type, and learning paradigm in similar works.
From: Multimodal fusion based few-shot network intrusion detection system
Model/method | K | Type | Dataset | ACC | Feat. Ext. | Data type | Learning paradigm |
|---|---|---|---|---|---|---|---|
Meta-learning (2022)38 | 10 | Multi | CICIDS2017 | 97.56% | CNN | Raw network traffic | Prototypical network |
FS-IDS (2022)39 | 5 | Binary | CICIDS2017 | 97.51% | CNN | Raw network traffic | Metric learning |
SPN (2023)40 | 5 | Binary | CICIDS2017 | 94.37% | CNN + Attention | Raw network traffic | Supervised learning |
GDE (2023)41 | 140 | Multi | CICIDS2018 | 99.13% | CNN | GAF image encoding | Diffusion model |
Siamese Network (2023)42 | 1 | Multi | CICIDS2017 | 80.81% | ANN | Statistical feature vectors | Siamese network |
MetaMRE (2023)43 | 10 | Binary | CICIDS2017 | 93.30% | Dilated causal conv | Raw network traffic | MAML |
MetaMRE (2023)43 | 10 | Multi | CICIDS2017 | 91.80% | Dilated causal conv | Raw network traffic | MAML |
FE-MTDM (2023)25 | 65341 (1%) | Multi | CICIDS2017 | 99.70% | Shallow NN + RF | Raw + statistical features | Prototypical network |
MAML+CNN (2023)3 | 5 | Multi | FSIDS IoT | 89.64% | CNN | Statistical feature vectors | MAML |
FML (2024)44 | 10 | Multi | CICIDS2017 | 87.27% | ResNet | Statistical feature vectors | Federated meta learning |
Res-Natural GAN (2024)45 | 15 | Binary | CICIDS2018 | 95.75% | GAN based CNN | Raw network traffic | Prototypical network |
Self-Sufficient Model | 5 | Multi | CICIDS2017 | 92.80% | CNN + Transformer | Raw + statistical features | Supervised learning |
Transfer-Enhanced Model | 5 | Multi | CICIDS2017 | 93.40% | CNN + Transformer | Raw + statistical features | Transfer learning |
Self-Sufficient Model | 10 | Multi | CICIDS2017 | 92.90% | CNN + Transformer | Raw + statistical features | Supervised learning |
Transfer-Enhanced Model | 10 | Multi | CICIDS2017 | 95.20% | CNN + Transformer | Raw + statistical features | Transfer learning |
Self-Sufficient Model | 5 | Multi | CICIDS2018 | 98.40% | CNN + Transformer | Raw + statistical features | Supervised learning |
Transfer-Enhanced Model | 5 | Multi | CICIDS2018 | 98.50% | CNN + Transformer | Raw + statistical features | Transfer learning |
Self-Sufficient Model | 10 | Multi | CICIDS2018 | 98.70% | CNN + Transformer | Raw + statistical features | Supervised learning |
Transfer-Enhanced Model | 10 | Multi | CICIDS2018 | 99.50% | CNN + Transformer | Raw + statistical features | Transfer learning |
