Toward Detecting Malware Based on Process-Aware Behaviors

Table 4

Performance of DNN model with various inputs.

Performance (%)	APIr	APIw	APIrs	APIr_Attr	APIw_Attr	APIws_Attr
(a) Validation results on DataSet1
Accuracy	97.89	97.96	98.05	98.01	98.09	98.21
Precision	98.51	98.43	98.39	98.35	98.48	98.70
Recall	97.25	97.48	97.71	97.67	97.69	98.71
F1-score	97.87	97.95	98.05	98.01	98.08	98.21
(b) Test results on DataSet2
Accuracy	91.09	94.58	94.72	93.15	96.78	97.63
Precision	91.91	94.83	94.85	93.58	96.82	97.64
Recall	91.09	94.58	94.72	93.15	96.78	97.63
F1-score	91.50	94.71	94.79	93.37	96.80	97.64

The data is bolded to indicate the best-performing result in terms of performance metrics.