Optimizing Dimensionality Reduction in SDN: A Metaheuristic Approach of UMAP Parameter Tuning

The reliability and performance of Intrusion Detection Systems (IDS) in Software-Defined Networking (SDN) can be significantly improved through effective dimensionality reduction. The Uniform Manifold Approximation and Projection (UMAP) is one of the most prominent methods for dimensionality reduction, making it ideal for structured and hierarchical network data. UMAP's performance is critically dependent on fine-tuned parameters such as the number of neighbors, minimum distance between points, and the distance metric, which directly impact the quality of reduced-dimensional space. This paper employs metaheuristic algorithms, including Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), and Simulated Annealing (SA) to methodically optimize these parameters. The application of these algorithms demonstrates significant improvements in the accuracy and reliability of UMAP, leading to more effective anomaly detection in SDN environments. This research not only enhances the operational efficiency of IDS but also contributes to the body of knowledge by providing a systematic approach to parameter tuning that can be applied to other AI-driven applications.