Hybrid Firefly and Particle Swarm Optimization designed for xgboost tuning problem: network intrusion detection

A Network Intrusion Detection System (NIDS) is a widely used tool for detecting attacks and securing networks, though it commonly faces the challenge of false positives. Based on a comparative analysis of Firefly algorithm for feature selection and Particle Swarm Optimization (PSO) and XGBoost, this paper introduces the FAPSO-XGBoost model, which achieves higher classification accuracy than alternative models such as XGBoost, Random Forest, Bagging, and AdaBoost. The approach begins by constructing a classification model using XGBoost, followed by PSO, which adaptively optimizes the structure of XGBoost. The proposed model is evaluated using the benchmark NSL-KDD dataset, and experimental results show that PSO-XGBoost outperforms other models in terms of precision and recall.