Network abnormal traffic detection method based on fusion of chord similarity and multiple loss encoder

EURASIP Journal on Wireless Communications and Networking

Table 6 Summary of the performance of each model on the NSL-KDD dataset

No.	Method	Accuracy (%)	Type
1	Multi-layer Preceptron [38]	77.41	Deep Learning based methods
2	Recurrent Neural Network [54]	83.28
3	AE+Guassian Naïve Bayes [59]	83.34
4	STL+SVM [60]	84.96
5	AE+SMR [61]	88.39
6	AE by Sadef et al. [51]	88.98
7	AE by Xu et al. [55]	90.61
8	DNN [51]	89.00
9	DCNN [62]	84.58
10	SVM [38]	69.52	Conventional classification methods
11	Naïve Bayes [38]	76.56
12	J48 [38]	81.05
13	NB tree [38]	82.02
14	Random forest [38]	80.67
15	Random tree [38]	81.59
16	Fuzzy approach [56]	84.12
17	NADLA	92.79