A Graph-Based Feature Generation Approach in Android Malware Detection with Machine Learning Techniques

Table 11

Comparison with the existing works.

Detection tool	Dataset		Feature set	Accuracy	Recall (%)
	Benign samples	Malicious samples
DroidMat [10]	1500	238	Component, system call, API, etc.	97.8	87.4
Drebin [5]	123453	5560	Component, permission, intent filter, sensitive API, website address, etc.	93.9	95.9
Literature [4]	5000	1260	API and permission.	91.52	94.23
Literature [6]	3000	7449	API and permission.	94.6	90.7
Our approach	2600	2130	Sensitive permission, bigram of API calls, and sensitive system broadcast.	95.4	96.5