A Hybrid Method Using Slime Mold Algorithm and Genetic Algorithm for Feature Selection Problems in Intrusion Detection Systems
ABSTRACT This paper presents an innovative hybrid approach for intrusion detection system (IDS) proposed by integrating the slime mold algorithm (SMA) and genetic algorithm (GA) within a feature selection (FS) framework for classification tasks. IDS faces challenges such as high‐dimensional data and...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-07-01
|
| Series: | Engineering Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/eng2.70254 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | ABSTRACT This paper presents an innovative hybrid approach for intrusion detection system (IDS) proposed by integrating the slime mold algorithm (SMA) and genetic algorithm (GA) within a feature selection (FS) framework for classification tasks. IDS faces challenges such as high‐dimensional data and the presence of irrelevant or redundant features, which can degrade detection accuracy and increase computational cost. To enhance search efficiency, opposition‐based learning (OBL) is utilized in the initialization phase, ensuring a well‐distributed initial population and accelerating convergence. While SMA exhibits strong exploration capabilities, its exploitation ability remains limited; therefore, GA is incorporated to reinforce exploitation and maintain a balance between exploration and exploitation. The proposed hybrid approach, OSMOGA, is applied for FS in ID problems and is rigorously evaluated against well‐established metaheuristic algorithms, including GA, grasshopper optimization algorithm (GOA), particle swarm optimization (PSO), teaching‐learning optimization (TLBO), and salp swarm optimization (SSA). Experimental results demonstrate that OSMOGA achieves superior detection accuracy while significantly reducing training time through effective FS. OSMOGA achieved classification accuracy of 98.64%, 98.99%, 99.43%, and 99.78% for NSL‐KDD, KDD Cup'99, CICIDS2017, and UNSW‐NB15 data sets, respectively. |
|---|---|
| ISSN: | 2577-8196 |