Performance Analysis of Classification Models for Network Anomaly Detection

The widespread use of internet-based services and products in the government and corporate sectors has resulted in a significant increase in individual internet usage. However, this move has also increased the susceptibility of systems to malicious activities. Recently, there has been significant interest in using deep learning methods to improve cybersecurity. This is because deep learning algorithms can effectively tackle important online security issues by employing advanced learning techniques. Machine learning (ML) and deep learning (DL) methodologies have been widely utilized in various aspects of cybersecurity, including vulnerability assessment, malware categorization, spam detection, and spoofing identification. A novel technique is proposed in this study for hierarchical intrusion detection. The proposed system detects the attack data and classifies the dataset into normal and attack (binary classification). The data is inputted into the initial steps and preprocessing step to prepare the data for the binary classifier to categorize it as either normal or an attack. In both classification frameworks, the data goes through important preprocessing steps, such as feature normalization, feature selection, making a Convolutional Neural Network (CNN) model with the KDD99 dataset, and then using the CNN model to find outliers.