TFDroid: Android Malware Detection by Topics and Sensitive Data Flows Using Machine Learning Techniques

With explosive growth of Android malware and due to the severity of its damages to smart phone users, efficient Android malware detection methods are urgently needed. As is known to us, different categories of applications divided by their functions use sensitive data in distinct ways. Besides, in e...

Full description

Saved in:
Bibliographic Details
Published in2019 IEEE 2nd International Conference on Information and Computer Technologies (ICICT) pp. 30 - 36
Main Authors Lou, Songhao, Cheng, Shaoyin, Huang, Jingjing, Jiang, Fan
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.03.2019
Subjects
Android malware detection
data flows
Data mining
Feature extraction
Games
Google
machine learning
Malware
Meteorology
Smart phones
topic modeling
Online AccessGet full text

Cover

More Information
Summary:With explosive growth of Android malware and due to the severity of its damages to smart phone users, efficient Android malware detection methods are urgently needed. As is known to us, different categories of applications divided by their functions use sensitive data in distinct ways. Besides, in each category, malicious applications treat sensitive data differently from benign applications. We thus propose TFDroid, a novel machine learning-based approach to detect malware using the related topics and data flows of Android applications. We test TFDroid on thousands of benign and malicious applications. The results show that TFDroid can correctly identify 93.7% of all malware.
DOI:10.1109/INFOCT.2019.8711179