malC: A Novel Deep Learning Architecture for Malware Classification

Malware attacks impact organizations resulting in password theft, encrypting the files, denial of service, etc. Every malware file has its own malicious intent, which either matches the known malware family or is detected as a new kind of malware. Analysts perform static and dynamic analysis on malware files to extract the behavior. Identifying the malware samples from the large data depends on the functionality and patterns. Feature extraction from the large data sets is mainly based on the expert’s domain knowledge and various approaches used to extract the important features from the raw input. Using deep-learning techniques, neural networks with multiple layers perform the feature extraction and classification altogether. In this paper, we design a novel convolutional neural network architecture named malC model identified as malware classification model. We applied the model on the benchmark dataset released for the Microsoft Malware Classification Challenge. The results show that the malC model outperforms several deep learning methods published in the literature.