Application of Multilayer Perceptron (MLP) neural network for detection and classification of cyber threats in network traffic

This article examines the application of a multilayer perceptron (MLP) for network traffic classification aimed at detecting cyber threats. The model was trained on the NSL-KDD dataset, a standard dataset widely used in research for attack detection tasks. During the experiments, data preprocessing was conducted, including encoding of categorical features and class balancing using the SMOTE method to address the imbalance between normal and malicious traffic. The results demonstrated high classification accuracy of 96,64 %, even under noise conditions and 10-fold cross-validation, which confirms the reliability of the proposed approach. The article presents performance metrics such as accuracy, recall, and F1-score, which can serve as a foundation for further research and optimization of machine learning models to enhance network security.