Alex/ELM Network Detection based on Improved Firefly Swarm Optimization Algorithm

Abstract

To address the issues of blind spots and low detection accuracy associated with using a single machine learning approach in network intrusion detection, the author suggests employing an Alex/ELM network detection system enhanced by an optimized firefly swarm algorithm. In the construction of base classifiers, the differences between samples of each base classifier are increased by sampling the sample set and selecting the feature set; By employing various learning algorithms to boost the diversity of the base classifiers on the sample set, the detection results are combined through a weighting mechanism. An enhanced firefly optimization algorithm is used to fine-tune the classification result weights of each base classifier. Experimental results demonstrate that, compared to other algorithms, this approach maintains a relatively high detection accuracy (with a minimum accuracy of 95.5%), showcasing the algorithm’s stability and effectiveness even with imbalanced samples. In conclusion, the method proposed by the author significantly enhances detection accuracy, while reducing both the false alarm rate and the missed alarm rate.