Distributed Swarm Learning for Edge Internet of Things

The rapid growth of Internet of Things (IoT) has led to Challenge-2: Non-convex optimization. Gradient-based algorithms the widespread deployment of smart IoT devices at wireless get trapped in local optima when tackling non-convex edge for collaborative machine learning tasks, ushering in a problems, e.g., training neural networks with nonlinear activation. With a huge number of hardwareconstrained This problem worsens in distributed learning, particularly IoT devices operating in resource-limited wireless in IoT scenarios where edge devices access limited data. Edge learning including communication and computation bottlenecks, device faces statistical heterogeneity in local training data across and data heterogeneity, security risks, privacy leakages, nonconvex workers, also known as the non-i.i.d. To heterogeneity in IoT hardware capability and link quality, address these issues, this article explores a novel framework which degrades edge learning performance significantly.