Evaluation of Deep Learning Methods Efficiency for Malicious and Benign System Calls Classification on the AWSCTD

Table 3

Malware-type classification with the help of DL and SVM methods (AllMalware set).

Count	Accuracy (percent)				Classification time (seconds)
	LSTM	GRU	CNN	SVM	LSTM	GRU	CNN	SVM
10	88.1	88.3	87.3	89.4	0.0000926	0.0000840	0.0000401	0.0000440
20	88.8	88.1	89.0	89.4	0.0001043	0.0000994	0.0000510	0.0000583
40	89.1	90.6	91.2	91.6	0.0001327	0.0001377	0.0000618	0.0000842
60	88.2	90.5	91.2	91.9	0.0001786	0.0001704	0.0000890	0.0000848
80	91.8	91.6	92.3	92.7	0.0002194	0.0002221	0.0001157	0.0000937
100	91.6	91.9	92.8	92.4	0.0002559	0.0002566	0.0001290	0.0001165
200	90.4	91.5	92.7	89.6	0.0004440	0.0004363	0.0003019	0.0002392
400	87.6	90.3	93.0	87.3	0.0009840	0.0008142	0.0006222	0.0006739
600	87.4	91.4	93.1	86.1	0.0023443	0.0023052	0.0016681	0.0015564
800	82.1	88.5	93.0	84.7	0.0043159	0.0037894	0.0023929	0.0022612
1000	75.5	89.6	93.1	83.2	0.0068075	0.0056159	0.0033276	0.0032245