N-Gram, Semantic-Based Neural Network for Mobile Malware Network Traffic Detection

Table 8

Description of detecting performance metrics of different deep learning architectures with different text sizes.

Deep learning type number	Evaluation metrics	Text size (byte) of the transformed data from network flow
		200	500	700	1000	1200	1500
1	Precision	0.671	0.672	0.682	0.713	0.718	0.722
	Recall	0.584	0.594	0.599	0.614	0.617	0.621
	F1-score	0.625	0.631	0.638	0.660	0.664	0.667
	Accuracy	0.625	0.633	0.640	0.658	0.661	0.665
2	Precision	0.711	0.761	0.827	0.843	0.869	0.878
	Recall	0.624	0.669	0.694	0.727	0.749	0.757
	F1-score	0.665	0.712	0.755	0.781	0.802	0.813
	Accuracy	0.666	0.713	0.749	0.779	0.800	0.812
3	Precision	0.768	0.817	0.883	0.916	0.914	0.936
	Recall	0.651	0.716	0.779	0.837	0.848	0.859
	F1-score	0.705	0.763	0.828	0.874	0.879	0.896
	Accuracy	0.701	0.764	0.829	0.878	0.883	0.899
4	Precision	0.767	0.816	0.883	0.915	0.908	0.934
	Recall	0.651	0.718	0.777	0.833	0.847	0.853
	F1-score	0.704	0.762	0.826	0.872	0.876	0.892
	Accuracy	0.700	0.765	0.827	0.875	0.881	0.895
5	Precision	0.847	0.894	0.954	0.969	0.971	0.971
	Recall	0.715	0.784	0.848	0.880	0.881	0.881
	F1-score	0.775	0.836	0.898	0.922	0.924	0.924
	Accuracy	0.772	0.836	0.899	0.924	0.925	0.926