Malware Elimination Impact on Dynamic Analysis: An Experimental Machine Learning Approach

Abstract

According to recent reports from security repositories, malware caused global resources to sustain losses equal to 11.7 million dollars during the last year. The expansion in the tendencies of the profiteers towards making use of malware is now being seen more intensified. The production of various tools has made it possible to produce and release malware with the least technical knowledge. In contrary, malware analysis tries preventing the expansion followed by the discovery of malware. Malware analysis can be divided into two main branches, namely static and dynamic analysis. Static analysis, for its limitations, like lack of program running, cannot be accountable alone to the discovery of new malware. Due to the same reason, dynamic analysis is currently being more widely applied and it is proved more reliable. One problem exists in the dynamic analysis is that the researches conducted in this regard eliminate many of the samples for such reasons as the corruption of the file or lack of proper running as well as some other reasons. This makes the results be unreliable in the real world because it is possible to infect the system by a malware like omitted instances. This chapter combines the static and dynamic analysis methods so that the problem of the eliminated samples could be solved. The proposed method has been able to improve the detection accuracy to 97%, with considering of the samples that have not been properly run.