AIML Based Deep Learning Network Intrusion Detection System

The growing complexity of cyber threats has made it essential to develop dependable network intrusion detection systems that can accurately identify both normal and harmful network traffic. This paper introduces an offline dataset-based Network Intrusion Detection and Prevention System (NIDS) built on a hybrid Convolutional Neural Network and Residual Feedforward Neural Network (CNN-RFNN) architecture designed for classifying network traffic. The proposed system operates on the CSE-CIC-IDS2018 dataset and handles network-flow records through a pipeline that includes data cleaning, feature extraction, normalization, and classification using 78 network traffic features. To assess how well the model performs on data it has not previously encountered, the dataset is split into 70% for training and 30% for testing, with the testing portion kept entirely separate during the training phase. The system combines the trained model with a security dashboard that offers attack classification, confidence levels, security alerts, threat analysis, and dataset-based simulation capabilities. Experimental findings show that the hybrid model is capable of reliably distinguishing normal traffic from various intrusion categories and delivering strong classification results on previously unseen network flows. The proposed framework offers a practical and reproducible method for integrating deep-learning-based intrusion detection with a security monitoring interface intended for network security analysis.