Classification of Malware using Deep Learning Techniques
Pooja Bagane, Susheel George Joseph, Abhishek Singh, Anurag Shrivastava, B. Prabha, Amit Shrivastava
Abstract
Malware remains a major threat, starting from home users to big business. That makes it a subject of hot study. Malware detection is achieved by means of static and dynamic study of malware signatures and activity patterns. These are shown to be ineffective and time consuming when unknown malware is being found. Many machine learning algorithms are created to recognize the new malware. Feature engineering is a crucial step in the construction of those algorithms. Which takes too long. This move can be wholly avoided by using deep learning techniques. Recent research has confirmed that many of them used skewed data collection, which in real-time circumstances is totally ineffective. Hence, this drives to build a new algorithm / architecture to use deep learning to detect malware. Using advanced Convolutional Neural Networks to identify patterns in malware sequences, using the weight sharing principle. We may catch recurring trends in malware by integrating this with Recurrent Neural Networks, too.