An Improved Binary Slime Mold Algorithm for Intrusion Detection Systems
Mahdieh Khorashadizade, Soodeh Hosseini, Morteza Jouyban · Concurrency and Computation Practice and Experience · 2025
ABSTRACT This paper proposes an enhanced intrusion detection system (IDS) that integrates an improved feature selection (FS) mechanism with optimized artificial neural network (ANN) training. The FS process is guided by a novel hybrid variant of the slime mold algorithm (SMA), called LBSMA, which incorporates both Lévy flight and Brownian motion to balance exploration and exploitation capabilities. Furthermore, an Equivalent SMA called ESMA is developed for training ANN by adopting the velocity update concept from the particle swarm optimization (PSO) algorithm. The proposed LBSMA‐ESMA framework is evaluated on several benchmark IDS data sets and compared with well‐known optimization techniques such as grasshopper optimization algorithm (GOA), PSO, genetic algorithm (GA), teaching‐learning optimization algorithm (TLBO), and Salp Swarm optimization algorithm (SSA). Experimental results show that the proposed method outperforms existing algorithms in terms of classification accuracy, convergence speed, and robustness, making it a promising solution for FS in security‐related applications.