Title: Poisoning-Resilient Federated Learning for Intrusion Detection in Industrial IoT

 

Speaker: Dr Segun Popoola

 

Abstract:
Federated learning is increasingly being explored for privacy-preserving intrusion detection in Industrial Internet of Things (IIoT) environments, where raw traffic data may be sensitive, distributed, or difficult to centralize. However, the same distributed training process creates a poisoning attack surface in which compromised clients can corrupt local data, flip labels, inject backdoor triggers, or submit malicious model updates that influence the global intrusion detection model. This presentation will provide a focused overview of poisoning-related threats in federated learning-based intrusion detection systems, including data poisoning, label-flipping attacks, backdoor poisoning, malicious updates, and Byzantine or outlying client behaviour. It will also critically review emerging defence strategies such as robust aggregation, anomaly filtering, trust-aware weighting, knowledge distillation, and backdoor-specific detection. The talk will conclude by presenting a conceptual framework for poisoning-resilient federated intrusion detection that combines client profiling, poisoning detection, robust aggregation, dynamic trust management, and recovery-oriented governance throughout the federated learning lifecycle.

 

Bio:
Dr Segun Popoola is a Senior Lecturer in the School of Computing and Information Science at Anglia Ruskin University, UK. His research focuses on the intersection of cyber security, artificial intelligence, federated learning, Internet of Things security, smart critical infrastructure, and wireless communications. He has published extensively in these areas, including work on federated deep learning for intrusion detection and zero-day botnet attack detection in IoT-edge environments. He is a Member of IEEE and has contributed to several research and innovation projects involving AI-driven cyber security, IoT security, and federated learning. His broader research interests include the security of AI systems, intrusion detection, deep learning, smart critical infrastructure, and privacy-preserving collaborative intelligence.