Abstract
Currently, the increasing number and complexity of cyberattacks make ensuring the security of computer networks one of the most critical challenges in modern information systems. Traditional signature-based methods are effective in detecting known attacks; however, their capabilities are limited when identifying new or modified attack types. Therefore, the use of intelligent methods for network traffic analysis becomes necessary. The aim of this study is to develop and evaluate an effective model for network attack detection using machine learning techniques. The CICIDS2018 dataset, which includes various types of cyberattacks and normal network traffic, was used as the experimental basis. The research involved data preprocessing, feature normalization, and the construction of an artificial neural network model for multiclass classification. The proposed model consists of two hidden layers and is trained using Dropout regularization and the Adam optimizer. Experimental results show that the model achieves an accuracy of approximately 96.5% and effectively detects different types of attacks. The findings demonstrate the potential of machine learning methods for improving intrusion detection systems and advancing modern cybersecurity technologies.