Convergence and Divergence Theories for Variational Decentralized Learning. Decentralized AI and learning meet the growing demand for hybrid, intelligent device, edge, and cloud systems and services.

Description

Convergence and Divergence Theories for Variational Decentralized Learning. Decentralized AI and learning meet the growing demand for hybrid, intelligent device, edge, and cloud systems and services. However, they face foundational challenges and knowledge gaps unexplored by existing learning systems. We aim to originate variational decentralized learning theories and methods to integrate variational, decentralized, and deep learning to satisfy complex stylistic, local-global integrative requirements. These transcend current aggregation-based learning frameworks by balancing local divergence and global convergence. The resulting groundbreaking theories and methods are foundational for real-world decentralized applications embedded with increasingly stylistic, divergent, and hierarchical settings and uncertainties.. Scheme: Discovery Projects. Field: 4602 - Artificial Intelligence. Lead: Prof Longbing Cao