MC-MLDCNN: Multichannel Multilayer Dilated Convolutional Neural Networks for Web Attack Detection

Table 9

The experimental results of MC-MLDCNN with various parameters.

Approach	Study or MC-MLDCNN with different parameters	Accuracy	Precision	Recall	score	FPR
MC-MLDCNN with different parameters	Model 1 (, , , )	99.12	99.39	98.46	98.93	0.42
	Model 2 (, , , , )	99.29	99.39	98.88	99.13	0.42
	Model 3 (, , , )	99.34	99.76	98.63	99.19	0.16
	Model 4 (, , , )	99.00	99.00	98.56	98.78	0.69
	Model 5 (, , , , )	99.34	99.62	98.75	99.19	0.26
	Model 6 (, , , , )	99.08	99.46	98.30	98.88	0.37
	Proposed (, , , )	99.36	99.65	98.80	99.22	0.24
Competitive deep learning models in the literature	Hao et al. [36]	98.35	99.00	98.17	98.58	1.40
	Jemal et al. [2]	99.25	97.73	99.35	98.53	—
	Gong et al. [19]	97.79	98.54	96.04	97.27	—
	Odumuyiwa and Chibueze [45]	96.39	98.83	95.0	96.88	2.00
	Rizvi et al. [52]	96.85	97.62	94.64	96.11	1.60
Classic deep learning models	CNN	97.93	97.96	96.97	97.46	1.40
	LSTM	97.70	97.91	96.46	97.18	1.43
	Bi-LSTM	97.54	97.98	96.00	96.98	1.38

The efficiency of the proposed methodology in comparison to benchmark deep learning models. The top two scores are in bold.