Enhancing Feature Selection with GMSMFO: A Global Optimization Algorithm for Machine Learning with Application to Intrusion Detection

Abstract The paper addresses the limitations of Moth-Flame Optimization (MFO) algorithm, a meta-heuristic used to solve optimization problems. MFO which employs moths' transverse orientation navigation technique, has been generate solutions for such However, performance is dependent on flame production and spiral search components, mechanism could still be improved concerning diversity flames ability find solutions. authors propose revised version called GMSMFO, uses Novel Gaussian mutation shrink enhance population balance exploration exploitation capabilities. study evaluates GMSMFO using CEC 2017 benchmark 20 datasets, including high-dimensional intrusion detection system dataset. proposed algorithm compared other advanced metaheuristics, its evaluated statistical tests as Friedman Wilcoxon rank-sum. shows that highly competitive frequently superior algorithms. It can identify ideal feature subset, improving classification accuracy reducing number features used. main contribution this research includes improvement exploration/exploitation expansion local search. ranging controller diversity. compares with traditional metaheuristic algorithms 29 benchmarks application binary selection benchmarks, systems. (Wilcoxon rank-sum Friedman) evaluate source code available at https://github.com/MohammedQaraad/GMSMFO-algorithm.