FWA-SVM Network Intrusion Identification Technology for Network Security

In the digital age, the increasing demand for network security has driven research on efficient network intrusion detection systems. The effectiveness of traditional network intrusion is limited in the face of complex network attacks and constantly increasing data volume. Support vector machine has attracted much attention for its excellent classification ability, but it faces challenges in feature selection and parameter optimization when dealing with large-scale high-dimensional data. Therefore, the study introduces the fireworks algorithm to improve it and optimize parameter selection and feature subset selection. The study also proposes a discretized binary fireworks algorithm to further improve the efficiency and adaptability of support vector machines in feature selection. The experiment outcomes denote that on the feature dense Sonar dataset, the average number of features selected by the proposed method is 24.39, a decrease of 25.51% compared to the comparison algorithm, and a classification accuracy improvement of 2.99%. The average detection rate of the raised method is 96.43%, the false alarm rate is 0.91, and the average correlation coefficient is as high as 0.987, which is better than the other four comparative algorithms. The training and testing time are 26.12 seconds and 11.23 seconds, respectively. Consequently, the primary contribution of the research lies in solving key problems in network intrusion detection through the BFWA-SVM model. Introducing the discrete fireworks algorithm to optimize support vector machines has improved the ability to process large-scale high-dimensional data. This model significantly reduces false positive and false negative rates, enhances real-time security awareness, and provides new guidance for network security management.