Work

Improving Transparency In Intrusion Detection Systems Through LIME And SHAP-Based Explanation Of MLP Models

Year: 2026
Type: article
Abstract: Intrusion Detection Systems (IDS) are critical for safeguarding network security byidentifying malicious activities in real-time. While Multi-Layer Perceptron (MLP) neuralnetworks have demonstrated high accuracy in detecting intrusions, their complex decisionmakingprocesses remain largely opaque, limiting their practical adoption. This studyexplores the application of Explainable Artificial Intelligence (XAI) techniques—specificallyLIME (Local Interpretable Model-agnostic Explanations) and SHAP (SHapley AdditiveexPlanations)—to enhance the interpretability of MLP-based IDS. By providing transparentand interpretable explanations of individual intrusion predictions, these methods help securityanalysts understand, trust, and effectively respond to alerts generated by the system. Thecomparative analysis highlights the strengths and limitations of LIME and SHAP in terms ofexplanation quality, computational efficiency, and applicability in real-world intrusiondetection scenarios. The integration of XAI with MLP models promises to bridge the gapbetween high-performance detection and explainability, advancing the development oftrustworthy cybersecurity solutions
Language: English
Open Access status: diamond
OpenAlex ID: W7217698440
FWCI:
Cites: 11
Cited by:
Related to: 10