Optimized Attention-based Long-short-term memory and Gated Recurrent Unit for Malware Detection in Windows

With the increasing usage of the internet, there are various threats to assuring the safety and security of the user's information in routine life. Among that, malicious software is considered the major one that is termed malware. Malware detection requires a sophisticated approach to minimizing the malware thread that breaks the computer operation. In recent times, malware writers help for neglecting detection based on the utilization of certain approaches like polymorphic, hiding, and also zero-day of the attack. On the other hand, commercial anti-spyware or anti-virus utilizes signature-based matching for identifying the malware for solving these kinds of attacks. To overcome this issue, we propose a new framework for malware detection with intelligent approaches. In this work, a new optimized deep learning is suggested for a malware detection approach in windows. Initially, the data collection is performed on standard windows malware data from the standard website. It is further fed to the data cleaning process. Next, the Fishers Discriminant Analysis (FDA) based feature is applied for getting the most significant features. Then, an efficient Long Short Term Memory-Gated Recurrent Unit (LSTM-GRU) is proposed for detecting the malware. The parameters present in the Attention-based LSTM-GRU are optimized by using the Elephant Herding Optimization (EHO) algorithm. Finally, the performance analysis is carried out and shows the efficient performance.