A Scalable Solution Partially Supervised Approach for Generation of Family Signatures against Android Malware

Abstract

Reducing the effort people need to combat malware is extremely practical. We portray a versatile, semi-administered structure to look at extensive datasets of Android applications and distinguish new malware families. Until 2010 the industry standard for the detection of pests. The applications are mainly based on signatures. Because every tiny change in malware makes them ineffective, often new signatures are created a task that requires a lot of time and resources from experienced experts. With the framework we suggest, applications can be automatically grouped into families and propose formal rules to identify them with 100% recall and fairly high accuracy. The families are either used to safely expand the expertise of experts on new samples or to reduce the number of applications requires thorough analysis. We have shown the effectiveness and scalability of the current approach Experiments in a database of 1.5 million Android applications. In 2018, the structure was effective sent on Koodous, a community oriented enemy of malware stage.