Knowledge Routing in Decentralized Learning

With mobile and Internet of Things devices becoming pervasive in our lives and recent advances in edge computational intelligence [e.g., edge artificial intelligence/machine learning (AI/ML)], it became evident that the traditional methods for training AI/ML models are becoming obsolete, especially with the growing concerns over privacy and security. This work highlights the key challenges that prevent edge AI/ML from seeing wide-range adoption in different sectors, especially for large-scale scenarios. We advocate a knowledge-centric design in which the produced knowledge in the network is discovered and routed to the knowledge requester. This work highlights the importance of knowledge routing for the proposed decentralized learning framework to be effective.