Application of a New Feature Generation Algorithm in Intrusion Detection System

Table 11

Comparison of two-classification test results.

Research	Method	Accuracy	Dataset
Al-Yaseen et al. [13] (2017)	SVM + ELM	95.75%	KDD99
Wang et al. [10] (2017)	Augmented Features + SVM	99.31%	NSL-KDD
Al-Qatf et al. [31] (2018)	SAE-SVM (binary classification)	99.41%	NSL-KDD training dataset with 10-fold cross-validation
Ömer Kasim [20] (2020)	Label encoding + normalization + AE-SVM	99.5%	NSL-KDD training dataset with 10-fold cross-validation
Thaseen, S [24] (2021)	Xgboost + MLP + BPN + LSTM	99%	UNSW_NB15
Proposed method	mgfg + Xgboost (binary classification)	100%	KDD99 test
Proposed method	mgfg + Xgboost (binary classification)	100%	KDD99 training dataset with 10-fold cross-validation
Proposed method	mgfg + Xgboost (five-classification)	100%	KDD99 test
Proposed method	mgfg + Xgboost (five-classification)	100%	KDD99 training dataset with 10-fold cross-validation
Proposed method	mgfg + Xgboost (binary classification)	100%	NSL-KDD test
Proposed method	mgfg + Xgboost (binary classification)	100%	NSL-KDD train dataset with 10-fold cross-validation
Proposed method	mgfg + Xgboost (five-classification)	100%	NSL-KDD test
Proposed method	mgfg + Xgboost (five-classification)	100%	NSL-KDD training dataset with 10-fold cross-validation
Proposed method	mgfg + Xgboost (binary classification)	100%	UNSW_NB15 test
Proposed method	mgfg + Xgboost (binary classification)	100%	UNSW_NB15 training dataset with 10-fold cross-validation
Proposed method	mgfg + Xgboost (ten-classification)	100%	UNSW_NB15 test
Proposed method	mgfg + Xgboost (ten-classification)	100%	UNSW_NB15 training dataset with 10-fold cross-validation
Proposed method	mgfg + Xgboost (three-classification)	100%	CSE-CIC-IDS2018 test
Proposed method	mgfg + Xgboost (three-classification)	100%	CSE-CIC-IDS2018 training dataset with 10-fold cross-validation