A Machine Learning Based Hierarchical Model for DDOS Attack Classification with Hyperparameter Optimization

Abstract

Distributed Denial-of-Service (DDoS) attacks are a major threat to modern distributed network environments. These attacks overwhelm network resources by exploiting vulnerabilities in system infrastructure, particularly targeting critical assets such as web servers and organizational network frameworks. This project proposes a deep learning–based approach for the detection and prediction of DDoS attacks in network traffic. Advanced deep learning techniques are employed to classify and identify malicious traffic patterns effectively. The proposed framework utilizes two classification models: Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM) networks. Before training the models, the dataset is preprocessed using the StandardScaler method to normalize the feature values and improve model performance. Deep learning techniques help in both the detection and classification of distributed denial-of-service attacks by learning complex patterns within network traffic data. The MLP model, a feed-forward artificial neural network, maps input features to corresponding output classes for attack identification. The LSTM classifier, which is well suited for sequential data analysis, is applied to recognize temporal patterns in network activity and classify abnormal events. To evaluate the effectiveness of the proposed system, a confusion matrix is generated to measure the accuracy and performance of the models in detecting potential threats. The entire framework is implemented using the Python programming language, providing an efficient and scalable solution for network fault detection and cybersecurity analysis.