Register to attend this P4 Developer Days webinar, “In-Network Inference with P4: From Stateless to Hybrid Approaches”
Date: January 21, 2026
Time: 8:00am Pacific
In-network machine learning (ML) techniques employ the P4 language to embed trained ML models directly into programmable network data planes. This has enabled novel applications in network security, routing optimization, and traffic classification, amongst others. This presentation charts the evolution of these techniques, starting with stateless, packet-level inference models like Henna. It then explores the shift to stateful, flow-level approaches as demonstrated in Flowrest. The talk culminates with Jewel, a novel hybrid system that performs joint packet and flow-level inference for improved accuracy and efficiency. This journey from stateless to hybrid methodologies highlights the advancements and trade-offs in building intelligent, high-performance, and ML-empowered networks with P4.
Aristide Akem is a Lecturer in Computer Science within the Cyberphysical Systems Group in the School of Electronics and Computer Science at the University of Southampton. Before joining Southampton, he was a postdoctoral researcher in the Computing Infrastructure Group at the University of Oxford. His research spans machine learning, network programming with P4, and mobile networking, with applications in network security, IoT, and energy. He earned his PhD from Universidad Carlos III de Madrid and IMDEA Networks Institute, following a Master of Science in Electrical and Computer Engineering from Carnegie Mellon University Africa, and a Master’s in Telecommunications Engineering from the University of Yaounde I.
