Binary File’s Visualization and Entropy Features Analysis Combined with Multiple Deep Learning Networks for Malware Classification

Table 3

The performance of byte-sequence-feature-based classifiers.

Classifier	Training time (s)	Acc (%)	F1-score (%)	AUC
RF	2.64	97.97	97.87	0.9965
KNN	1.53	98.35	98.35	0.9890
MLP	157.90	90.36	88.91	0.9928
SVM	61.90	98.87	98.84	0.9996
DT	14.41	96.47	98.84	0.9559
NB	0.226	96.83	96.97	0.9813

The significance of the bold values given in the table is that, in the experiment, the SVM classifier achieved the best classification performance, and the classification accuracy rate can reach 98.87%.