Malware Detection with Neural Network Using Combined Features
Huan Zhou · Communications in computer and information science · 2019
The growth in amount and species of malicious programs are now turning into a severe problem that strengthens the demand for development in detecting and classifying the potential threats automatically. Deep learning is an acceptable method to process this increment. In this paper, we propose an innovative method for detecting malware which uses the combined features (static + dynamic) to classify whether a portable executable (PE) file is malicious or not. A thorough experimental research on a real PE file collection was executed to make comparisons with the results that was performed in diverse situations and the performances of different machine learning models. The experiments prove the effectiveness of our model and show that our method is able to detect unknown malicious samples well.