Hybrid Feature Hashing and Recurrent Neural Network based LSTM Architecture for Real Time Malware Analysis and Detection Techniques

A website is designed based on a python framework for real-time malware detection. The proposed malware detection techniques categorize the file as legitimate or malicious and also display the relevant webpage. The important features of the uploaded file are predicted and analyzed by employing the advantage of feature hashing techniques and RNN based LSTM architecture. Here, initially convert the categorical to numerical values and then a complete dataset is used for extracting important features. Feature hashing is a strong technique to deal with high-dimensional features. RNN based LSTM architecture and Extreme Gradient Boosting is used to train the model and obtain an accuracy of 98.5% each respectively, and LSTM gave 98.9%. As a result, in approximation, the model would predict 99 correct results out of 100 uploaded files. Hence, most malware files would be detected by our proposed algorithm and can be a good fit for real-time malware detection.