Abstract: Current antivirus software’s are effective against known viruses, if a malware with new signature is introduced then it will be difficult to detect that it is malicious. Signature-based detection is not that effective during zero-day attacks. Till the signature is created for new (unseen) malware, distributed to the systems and added to the anti-malware database, the systems can be exploited by that malware. But Machine learning methods can be used to create more effective antimalware software which is capable of detecting previously unknown malware, zero-day attack etc. We propose an approach that learns from the header data of PE32 files. We examine various features of the PE32 header and check those which are suitable for machine learning classifier. We hypothesize that machine learning classifiers can tell apart the difference between malware and benign software. Various machine learning methods such as Support Vector Machine (SVM), Decision tree, Logistic Regression and Naive Bayes will be used
Keywords: Malware, detection, Feature extraction, machine learning, Classifier, SVM, Decision Tree, Naïve Bayes, Header Data.