A Survey on Collaborative Learning for Intelligent Autonomous Systems

Abstract

This survey examines approaches to promote Collaborative Learning in distributed systems for emergent Intelligent Autonomous Systems (IAS). The study involves a literature review of Intelligent Autonomous Systems based on Collaborative Learning, analyzing aspects in four dimensions: computing environment, performance concerns, system management, and privacy concerns, mapping the significant requirements of systems to the emerging Artificial intelligence models. Furthermore, the paper explores Collaborative Learning Taxonomy for IAS to demonstrate the correlation between IoT, Big Data, and Human-in-the-Loop. Several technological open issues exist in the aforementioned domains (such as in applications of autonomous driving, robotics in healthcare, cyber security, and others) to effectively achieve the future deployment of Intelligent Autonomous Systems (IAS). This Survey aims to organize concepts around IAS, indicating the approaches used to extract knowledge from data in Collaborative Learning for IAS, and identifying open issues. Moreover, it presents a guide to overcoming the existing challenges in decision-making mechanisms with IAS, providing a holistic vision of Big Data and Human-in-the-Loop.

FedRDS: Federated Learning on Non-IID Data via Regularization and Data Sharing

“…Wang et al [34] considered computation heterogeneity as the bottleneck of FL and therefore used two near-optimal algorithms to schedule the IID vs. non-IID data workloads. The collaborative learning classification method in the IAS decision system plays an important role in resource management and performance optimization in heterogeneous network environments [35].…”