Research Article

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

Table 1

Training and testing sets labelling.

Set labelSequence variationsComments

AllMalware10, 20, 40, 60, 80, 100, 200, 400, 600, 800, 1000Only malware samples
AllMalware210, 20, 40, 60, 80, 100, 200, 400, 600, 800, 1000Only malware samples with no more than two identical sequences in repetition
AllMalwarePlusClean10, 20, 40, 60, 80, 100, 200, 400, 600, 800, 1000Malware samples plus and benign samples as additional class
AllMalwarePlusClean210, 20, 40, 60, 80, 100, 200, 400, 600, 800, 1000Malware samples and benign samples as an additional class with no more than two identical sequences in repetition
MalwarePlusClean10, 20, 40, 60, 80, 100, 200, 400, 600, 800, 1000Only two classes to train: malware and benign
MalwarePlusClean210, 20, 40, 60, 80, 100, 200, 400, 600, 800, 1000Only two classes to train: malware and benign samples with no more than two identical sequences in repetition