NETWORK INTRUSION DETECTION SYSTEM BASED ON CNN AND DATA BALANCING

Abstract

The developing danger of digital attacks accentuates the need major areas of strength for security. Our Network Intrusion Detection (NID) arrangement utilizes Convolutional Neural Networks (CNNs) to address uneven datasets and further develop classification accuracy. To adjust order among assault types, the framework utilizes methods like ROS, Destroyed, and ADASYN to address information lopsided characteristics. NSL-KDD and BoT-IoT benchmark datasets show the framework's classification accuracy in recognizing and classifying network intrusions. This study adds ensemble draws near, for example, CNN and LSTM networks, to the first model's prosperity. This hybrid ensemble model accomplishes close to 100% accuracy. The ensemble method further develops framework strength and execution by pooling model predictions. This exploration accentuates network security and offers down to earth exhortation on utilizing complex deep learning calculations to further develop intrusion detection systems.[19]