A Robust Long Short-Term Memory Model for Classification of Malware Analysis

Malicious software (malware) analysis and classification has obtained a greater development in the systems connected by the Internet. The malware accomplishes the system's data and withdraws significant user data without any declaration. Furthermore, the malware secretly directs the data to the servers which are standardized by the attackers. Recently, various scientists and researchers identified the products of anti-malware to detect known malware. However, these approaches are insufficient to detect packed and obfuscated malware. To solve these issues, Cuckoo Search Algorithm (CSA) was developed for the classification of malware. The efficient features were chosen by the CSA approach which simplifies the process to classify the malware. The classifier of Long Short- Term Memory (LSTM) was employed in this research to classify the malware. The samples of malware are obtained from the VXHeavens dataset which includes malware samples from numerous software. The developed model attained better accuracy of 99.03%, sensitivity of 98.16%, recall of 98.25%, and fl-score of 97.11 % respectively than the existing models.