ODL: Opportunistic Distributed Learning for Intelligent IoT Systems

Abstract

As we transition from centralized machine learning to distributed learning, new practices can significantly enhance intelligent Internet of Things (IoT) systems. This article introduces the concept of Opportunistic Distributed Learning (ODL), a general framework that enables any node in a network to initiates learning tasks by leveraging local, unused distributed resources collaboratively. ODL, facilitated by edge intelligence, promotes collective responsibility, pervasive and flexible distributed learning, allowing participating nodes to freely move, group, and regroup based on their conditions and benefits. The article discusses key research challenges of ODL in intelligent IoT systems, presents the ODL framework, proposes a reputation-based node selection scheme, and highlights the benefits and future research directions of the ODL system.