Monitoring the behavior of program execution at run-time is widely used to differentiate benign and malicious processes executing in the host computer. Most of the existing run-time malware detection methods use the information available in Windows Application Programming Interface (API) calls. The proposed malware detection system uses the Windows API call sequence. A 3rd order Markov chain (i. e. 4-grams) is used to model the API calls. This composite feature set is provided as an input to the malware detection system to raise the final alarm. Association mining based classification is used because it yields higher detection accuracy than previous data mining based detection systems which employed Naive Bayes, Support Vector Machine and Decision Tree techniques. A minimal subset of API categories is monitored while maintaining high detection accuracy. The number of generated rules is reduced, by removing the redundant rules, to make the malware analysis efficient. The key novelty of the proposed malware detection system is the iterative learning process combined with the run-time monitoring of program execution behavior which makes this as a dynamic malware detection system. The performance of the proposed malware detection system is evaluated for accuracy of malware detection system and compared with the existing data mining based detection systems. It is inferred that the proposed malware detection system outperforms the existing malware detection systems.

Malware Detection Windows Api Calls Machine Learning