Skip to main content

Table 6 Comparison of classification rates with existing approaches in the literature on UNSW-NB15 data-set

From: Intrusion detection in internet of things using supervised machine learning based on application and transport layer features using UNSW-NB15 data-set

References Data used Technique Imputation Classification accuracy (%)
Binary Multi-class Multi-class clusters
Flow features Transport features Top features
[8] 10% CNN-1D 89.80 78.20
RF 87.90 73.20
SVM (Linear) 84.60 65.20 -
MLP 86.60 74.90
[9] 30% DT 95.32 (DNS) 97.13 (HTTP) -
NB 91.17 (DNS) 95.91 (HTTP)
ANN 92.61 (DNS) 96.27 (HTTP)
Ensemble 99.54 (DNS) 98.97 (HTTP)
[10] 10% GBT 93.13
KNN 91.90
DT 92.29
LR 92.35
NB 92.52
SVM 92.32
[11] 10% DT 89.86 (22 Features)
Proposed approach 60% RF Mean, multiple and linear regression 98.67 97.37 96.96 91.40 97.54
SVM   97.69 95.67 89.78 82.96 89.93
ANN   94.78 91.67 86.37 81.63 87.68
  1. Bold values show highest accuracies achieved by proposed solution
\