AI-Based Intrusion Detection System Enhanced with Swarm Intelligence
| dc.contributor.author | Abd-el-ali; Bekhouche | |
| dc.date.accessioned | 2026-09-27T16:17:32Z | |
| dc.date.available | 2026-09-27T16:17:32Z | |
| dc.date.issued | 2026 | |
| dc.description.abstract | Intrusion Detection Systems (IDS) are essential components of modern network securityinfrastructure. Thisthesispresentsthedesignandimplementationofanintelligent IDS that combines machine learning with swarm intelligence optimization. The system uses Random Forest as the base classifier and Particle Swarm Optimization (PSO) to automatically tune hyperparameters including the number of estimators, tree depth, minimum samples per split, and classification threshold. The system is evaluated on two standard benchmark datasets—NSL-KDD and CICIDS2017—using rigorous evaluation protocols. On NSL-KDD, PSO optimization improved the Detection Rate from 64.52% to 69.09% (+4.57%) while maintaining 96.96% precision. On CICIDS2017 with a stratified 80/20 split, PSO improved the Detection Rate from 93.89%to99.72%(+5.83%)bydiscoveringoptimaltreestructure, classbalancing, and threshold configuration. These results demonstrate the significant practical value of swarm intelligence for automated IDS hyperparameter optimization. Intrusion Detection Systems (IDS) are essential components of modern network securityinfrastructure. Thisthesispresentsthedesignandimplementationofanintelligent IDS that combines machine learning with swarm intelligence optimization. The system uses Random Forest as the base classifier and Particle Swarm Optimization (PSO) to automatically tune hyperparameters including the number of estimators, tree depth, minimum samples per split, and classification threshold. The system is evaluated on two standard benchmark datasets—NSL-KDD and CICIDS2017—using rigorous evaluation protocols. On NSL-KDD, PSO optimization improved the Detection Rate from 64.52% to 69.09% (+4.57%) while maintaining 96.96% precision. On CICIDS2017 with a stratified 80/20 split, PSO improved the Detection Rate from 93.89%to99.72%(+5.83%)bydiscoveringoptimaltreestructure, classbalancing, and threshold configuration. These results demonstrate the significant practical value of swarm intelligence for automated IDS hyperparameter optimization. Keywords: Artificial Intelligence, Intrusion Detection System, Swarm Intelligence, Machine Learning, Cybersecurity, Particle Swarm Optimization, Random Forest | |
| dc.identifier.uri | http://dspace.univ-khenchela.dz:4000/handle/123456789/11377 | |
| dc.language.iso | en | |
| dc.title | AI-Based Intrusion Detection System Enhanced with Swarm Intelligence | |
| dc.type | Thesis |