New FBID framework boosts IoT attack detection using adaptive PFL

Researchers have developed FBID, a new adaptive Personalized Federated Learning (PFL) framework designed to enhance out-of-distribution (OOD) attack detection in Internet of Things (IoT) networks. Unlike existing PFL methods that can lead to over-personalization and degraded OOD detection, FBID uses server-side control. It incorporates a contextual multi-armed bandit to dynamically adjust client training intensity and a trust-based blending mechanism to balance global and local model knowledge. Experiments on the CICIoT2023 dataset demonstrated FBID's ability to improve OOD Detection Rate and F1-Score, particularly against unseen attack classes. AI

IMPACT Enhances security in IoT networks by improving the detection of novel and evolving cyber threats.

RANK_REASON Academic paper detailing a new method for AI application. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• CICIoT2023
• Internet of Things

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →