Deep Neural Network-based Method for Detection and Classification of Malicious Network Traffic

Today's network intrusion threats are diverse and ever-increasing. The weaknesses and vulnerabilities of software, hardware, network protocols, and network management cause such types of threats. Intruders and criminals have now mastered the new methods to target and circumvent the legitimate ways to damage the devices with the exponential rise in technology. Now the stronger Intrusion Detection System (IDS) are required to classify network traffic with the best performance. We suggest a Deep Neural Network (DNN) based IDS in this study, which tracks traffic from both authorized and non-authorized sources. It reliably defines and extracts malicious traffic with a 99.79 percent accuracy for KDD-99 and 96.53 percent accuracy for NSL-KDD datasets. This study offers an updated version of the Deep Neural Network to solve design algorithm and over-fitting problem, and an evaluation has also been made with current models to show the efficacy, proficiency, and performance of the planned model.