Cybersecurity has emerged as a substantial hurdle in contemporary society. The rapid advancement of technology has led to an unprecedented reliance on new technologies. This reliance has not only elevated living standards and facilitated civilization but has also attracted individuals who continuously devise sophisticated attacking methods. As a consequence, the identification, prevention, and mitigation of security risks become challenging, requiring swift and efficient management. The rising frequency and appearance of new cyber-attacks necessitates the continual updating of threat detection algorithms. While traditional signature-based techniques are effective against typical attacks, they are unable to detect unknown or novel ones. As a result, intrusion detection systems (IDS) that rely on machine learning (ML) are growing in popularity. However, it takes a very long time to recognize a cyber-attack when a lot of features are present. The current research uses fuzzy-rough sets to provide an intelligent system for reducing the dimensionality of such data sets. By maximizing each feature’s relevance and significance, this approach aims to eliminate redundancy from the reduced feature set. In order to demonstrate the effectiveness of the proposed technique on a real-world data set and to compare it with other detection strategies, comprehensive results are provided using the widely used classifiers.