Table 7 The performance of LLMs based-data augmentation on the NF-ToN-IoT-v2 dataset.

	Our	CTGAN	SMOTE	ADASYN	Oversample	Undersample	Original
Benign	0.983	0.975	0.982	0.975	0.984	0.968	0.982
Scanning	0.987	0.986	0.990	0.982	0.989	0.976	0.986
XSS	0.944	0.945	0.946	0.878	0.946	0.938	0.944
DDoS	0.970	0.969	0.972	0.941	0.972	0.960	0.967
Password	0.915	0.914	0.921	0.878	0.922	0.907	0.915
DoS	0.903	0.903	0.904	0.482	0.903	0.901	0.902
Injection	0.822	0.820	0.830	0.775	0.835	0.785	0.822
Backdoor	0.995	0.386	0.986	0.977	0.984	0.972	0.992
MITM	0.117	0.035	0.088	0.032	0.102	0.073	0.064
Ransomware	0.903	0.932	0.881	0.870	0.682	0.565	0.033
F1-Macro	0.854	0.786	0.850	0.779	0.832	0.804	0.761
Accuracy	0.958	0.953	0.957	0.902	0.957	0.945	0.956

Quick links

Search