KTSDroid: A Framework for Android Malware Categorization Using the Kernel Task Structure

Table 2

Related work on malware detection using kernel task structure features.

Study	Number of effective features	Feature selection techniques	Algorithm classification	Reported performance	Data set size	Multiclass classification	KTS category analysis
Wang and Li [12]	10–40 out of 112	PCA, correlation, IG, and chi-square	Naïve Bayes, decision trees, and neural network	ACC: 94%–98%	1275 malware, 1275 benign		Partial
Alawneh et al. [14]	43 out of 112	Logistic regression	Neural network	ACC: 96.80%	1200 malware, 1200 benign
Shahzad et al. [15]	32 out of 90	Correlation	Decision trees (J48)	ACC: 93%–96%	110 malicious, 110 benign
Kim and Choi [20]	36 out of 59	Manual	Support vector machines (SVM)	ACC: 98.85%,	6 malicious