Artificial Intelligence and Machine Learning in Cyber Security: A Deep Dive Into State-of-the-Art Techniques and Future Paradigms

As malware's hazards grow from home computers to industrial control systems, detection has become mission crucial. In nowadays, cyber security task became an important and competitive work to secure the networks and systems from different attacks that are happening digitally. So for this we proposed a project which can reduce some of the attacks. This project describes malware detection system predicated upon deep learning for detecting as well as categorizing malevolent software. Our article offers convolutional neural network-based malware detection method that is very accurate and efficient. The system proceeds with binary file as input and determines whether it’s harmful or benign. The binaries are just lightly pre-processed, and feature discovery is left to the network during training. A key distinction from existing convolutional neural networks is used in this proposed method. Unlike previous existing here we use ember dataset as it contains of thousands of malware and benign files. The existing methods are using maligma dataset which is very small, in the view of getting accurate results while training. As the part of research we noted the results of existing models that are up to 75-82% accuracy. As they are using small datasets and models like knn, decision trees and as well as SVM, we are using different method and algorithm called CNN and got the better results than previous existing methods. That is 85-90% accuracy. We did this project on VM, because we are testing the model with original malware so that system is not affected.