Effective Malware Detection Based on Behaviour and Data Features

摘要: Malware is one of the most serious security threats on Internet today. Traditional detection methods become ineffective as malware continues to evolve. Recently, various machine learning approaches have been proposed for detecting malware. However, either they focused behaviour information, leaving data information out consideration, or did not consider too much about new with different behaviours versions obtained by obfuscation techniques. In this paper, we propose an effective approach using learning. Different from existing work, take into account only but also namely, opcodes, types and system libraries used in executables. We employ our implementation. Several experiments are conducted evaluate approach. The results show that (1) classifier trained Random Forest performs best accuracy 0.9788 AUC 0.9959; (2) all features (including types) detection; (3) capable some fresh malware; (4) has a resistance