SVM Based Effective Malware Detection System

Abstract—Malware is coined as an instance of malicious code that has the potential to harm a computer or network. Recent years have encountered massive growth in malwares as existing signature based malware detection approaches are becoming ineffective and intractable. Cyber criminals and malware developers have adapted code obfuscation techniques which undermines the effectiveness of malware defense mechanism. Hence we propounded a system which focuses on static analysis in addition with automated behavior analysis in emulated environment generating behavior reports to investigate malwares. The proposed method uses programs as opcode density histograms and reduces the explosion of features. We employed eigen vector subspace analysis to filter and diminish the misclassification and interference of features. Our system uses a hybrid approach for discovering malware based on support vector machine classifier so that potential of malware detection system can be leveraged to combat with diverse forms of malwares while attaining high accuracy and low false alarms