LENS-XAI, an intrusion detection framework for Industrial IoT that unifies knowledge distillation (compressing a strong teacher into a lightweight student for resource-constrained devices), a variational autoencoder (learning compact representations for anomaly-based detection), and attribution-based explainability (making each detection decision interpretable). Trained on only 10% of available data for efficiency, it is evaluated on four benchmark datasets (Edge-IIoTset, UKM-IDS20, CTU-13, NSL-KDD) with high accuracy and reduced false positives, targeting scalable and transparent network security.
LENS-XAI is a lightweight, explainable intrusion detection framework for Industrial IoT (IIoT) that unifies three ingredients: knowledge distillation to compress a strong teacher model into a compact…