Network Traffic Classification Based on SD Sampling and Hierarchical Ensemble Learning

Table 2

Related works of improved machine learning algorithms.

Detailed method	Literature	Description	Dataset	Best accuracy (%)
Ensemble learning	[13]	Ensemble voting based on classifier resolution and a multitree ensemble model	KDDTest+	85.20
	[14]	An AdaBoost model combining IForest, LOF, K-means algorithms	Datasets in UCI machine learning library	96.29
	[16]	A multiclassifier ensemble algorithm based on probability weighted voting	NSL-KDD	95.70
	[17]	A weighted majority algorithm based on the random forest	NSL-KDD	90.48
	[18]	A multilayer random forest model based on category detection and a partition technique	KDD Cup 1999	94.36
	[19]	An incremental learning framework	CICIDS2017	86.70
	[20]	A framework based on hard/soft combinators	Real traffic data	79.20
	[21]	Investigate and evaluate voting, stacking, bagging, boosting ensemble frameworks	Datasets in UCI machine learning library	99.97
Model optimization	[22]	A multidimensional stochastic hyperplane isolation method	Personal real data	AUC = 100.00
	[23]	The stacked LSTM model combines the idea of sliding windows	Personal real data	AUC = 91.55
	[24]	A multimodal data allocation framework MIMETIC	Real traffic data	96.74
Sample optimization	[25]	Naive Bayes feature embedding method	UNSW-NB15/CICIDS2017	93.75/98.92
	[18]	An outlier detection algorithm based on KNN	KDD Cup 1999	94.36