Explainable Artificial Intelligence with Single Layer Feedforward Neural Network and Improved Crowned Porcupine Optimization Algorithm for Classification Problems

The increasing occurrence of network intrusions calls for the development of advanced Artificial Intelligence (AI) techniques to tackle classification challenges in Intrusion Detection Systems (IDSs). However, the complex decision-making processes of AI often prevent human security professionals from fully understanding the behavior of the model. Explainable AI (XAI) enhances trust in IDSs by providing transparency and assisting professionals in interpreting data and reasoning. This study explores AI techniques that improve both accuracy and interpretability, strengthening trust management in cybersecurity. Integrating performance with explainability improves decision-making and builds confidence in automated systems for classifying network intrusions. This study presents an Explainable Artificial Intelligence Kernel Extreme Learning Machine Improved with the Crowned Porcupine Optimization Algorithm (XAIKELM-ICPOA) approach. Initially, the proposed XAIKELM-ICPOA method preprocesses the data using min-max scaling to ensure uniformity and improve model performance. Next, the Kernel Extreme Learning Machine (KELM) model is employed for classification. The Improved Crowned Porcupine Optimization (ICPO) method is used to optimize KELM hyperparameters, improving classification performance. Finally, SHAP is employed as an XAI technique to provide insights into feature contributions and decision-making processes. The XAIKELM-ICPOA method was evaluated on the NSL-KDD dataset, achieving an accuracy of 96.82%.