A Malware Detection Scheme via Smart Memory Forensics for Windows Devices

Table 1

Comparison of detection performance of DCNN model on three types of feature sets.

Feature	Dimension	Accuracy	Sample	Precision	Recall	F1-score
LBP	224 × 224	0.9297	Malware	0.90	0.61	0.73
			Benign	0.93	0.99	0.96
	300 × 300	0.9399	Malware	0.97	0.96	0.96
			Benign	0.80	0.82	0.81
GLCM	224 × 224	0.8873	Malware	0.95	0.28	0.43
			Benign	0.88	1.00	0.94
	300 × 300	0.8929	Malware	0.87	0.36	0.50
			Benign	0.89	0.99	0.94
LBP + GLCM	224 × 224	0.9780	Malware	0.97	0.98	0.98
			Benign	0.98	0.97	0.98
	300 × 300	0.9807	Malware	0.98	0.99	0.98
			Benign	0.99	0.98	0.98