An Efficient Cyber Security Framework for Network Intrusion Detection using Hybrid Classifier

The recent enormous increase in data volume and its ongoing growth has considerably increased the importance of information security and data analysis systems within the field of Big Data. An Intrusion Detection System (IDS) is critical in detecting unauthorized access or breaches within a system or network by actively monitoring and scrutinizing data. However, the large volume, diversity, and rapid rate at which networks generate data have made it difficult for traditional intrusion detection technologies to identify attacks efficiently. This paper focuses on applying machine learning techniques, specifically the Multi-Layer Perceptron (MLP) classifier, for network intrusion detection. The work begins with data exploration and preprocessing, removing unnecessary features and encoding categorical attributes. Numerical attributes are scaled using standardization techniques to ensure compatibility. Feature selection methods, using a random forest-based classifier to distribute feature importance, are employed to optimize the IDS performance. The selected features are used to train an MLP classifier suitable for learning complex patterns and making accurate predictions. The system is validated on a separate test dataset to assess its generalization and effectiveness. The results with 99% accuracy demonstrate the MLP classifier’s effectiveness in identifying network intrusions and detecting anomalous activities. The paper contributes to network security by showcasing the potential of machine learning techniques, specifically the MLP classifier, in developing robust and efficient IDS.