Table 5.
Analysis of ML-Based NIDS in IoT.
| Reference | ML Methods | Precision | Recall | Accuracy |
|---|---|---|---|---|
| Mayhew et al. [189] | Behavior-Based Access Control | 99.1% | - | - |
| Haddadpajouh et al. [174] | RNN, LSTM | - | - | 98.18% |
| Saeed et al [175] | RNN | - | - | 97.23% |
| Azmoodeh et. al. [176] | Deep Eigenspace learning | 98.59% | 98.37% | 99.68% |
| Erfani et al [177] | One-Class SVM | - | - | - |
| La et al. [179] | Bayesian Game Theory | - | - | - |
| Arrington et al. [181] | Behavioral Modeling | - | - | - |
| Li et al. [190] | KNN | 98.5% | - | - |
| Pajouh et al. [182] | Naïve Bayes | 84.86% | - | - |
| Ghosh and Mitra. [183] | Logistic Regression | - | - | 93.26% |
| Prokofiev et al. [184] | Logistic Regression | 94.0% | 98.0% | 97.30% |
| Singh and Neetesh [191] | Self-Organizing Map | 64% | - | - |
The “Applicable with Edge” column was established based on our observation on various system implementation.