Abstract:Federated learning (FL) and split learning (SL) are two emerging collaborative learning methods that may greatly facilitate ubiquitous intelligence in Internet of Things (IoT). Federated learning enables machine learning (ML) models locally trained using private data to be aggregated into a global model. Split learning allows different portions of an ML model to be collaboratively trained on different workers in a learning framework. Federated learning and split learning, each has unique advantages and respective limitations, may complement each other toward ubiquitous intelligence in IoT. Therefore, combination of federated learning and split learning recently became an active research area attracting extensive interest. In this article, we review the latest developments in federated learning and split learning and present a survey on the state-of-the-art technologies for combining these two learning methods in an edge computing-based IoT environment. We also identify some open problems and discuss possible directions for future research in this area with a hope to further arouse the research community's interest in this emerging field.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: How to combine Federated Learning (FL) and Split Learning (SL) in the Internet of Things (IoT) environment to achieve Ubiquitous Intelligence. Specifically, the paper focuses on how to overcome the challenges in achieving ubiquitous intelligence in the edge - computing - supported IoT environment by combining these two collaborative learning methods. ### Problem Background 1. **Data Scatter** : - The amount of data generated by IoT devices is huge and highly scattered. Traditional centralized cloud infrastructures cannot meet the requirements for real - time performance and privacy protection. - Transmitting data to the cloud not only consumes bandwidth but may also introduce additional latency, and data related to user privacy needs to be retained on local devices. 2. **Resource Limitations** : - Edge devices usually have limited computing and network resources and it is difficult for them to independently complete the training of complex machine - learning models, especially complex models such as deep neural networks. ### Solutions - **Federated Learning (FL)** : - It allows multiple data owners to collaboratively train a global model without exposing their private data. Each client uses local data to train a local model and then uploads the model parameters to the server for aggregation. - Advantage: Protects data privacy and reduces communication overhead. - Limitation: It requires client devices to have sufficient resources to train a complete model, which is a challenge for resource - limited IoT devices. - **Split Learning (SL)** : - It divides a machine - learning model into multiple parts, and different parts can be collaboratively trained on different nodes. The client executes the front - end part of the model and sends the intermediate results to the server, and the server continues the training of the back - end part and returns the gradients. - Advantage: Allows the client to offload part of the training tasks to the server, making full use of the flexible resource management in edge computing. - Limitation: The sequential collaboration between the client and the server limits the effective use of large - scale distributed data. ### Advantages of Combining FL and SL - **Complementary Advantages** : - FL solves the problems of data privacy and communication efficiency, while SL alleviates the resource limitation problem. The combination of the two can give full play to their respective advantages and at the same time make up for their respective deficiencies. - For example, on resource - limited IoT devices, SL can be used to offload computationally intensive tasks to edge servers, and in terms of data privacy protection, FL can be relied on to ensure that data does not leave the local area. ### Main Contributions of the Paper - **Latest Progress Review** : - This article reviews the latest developments of FL and SL and provides a survey of the current state of technology on how to combine these two methods. - **Open Problems and Future Directions** : - It identifies some unsolved problems and discusses possible directions for future research, aiming to further stimulate the research community's interest in this emerging field. In this way, the paper provides researchers with a comprehensive perspective to help them understand how to achieve more efficient ubiquitous intelligence in the edge - computing - supported IoT environment.

Combined Federated and Split Learning in Edge Computing for Ubiquitous Intelligence in Internet of Things: State of the Art and Future Directions

Combining Federated Learning and Edge Computing toward Ubiquitous Intelligence: Challenges, Recent Advances, and Future Directions

Federated Split Learning for Edge Intelligence in Resource-Constrained Wireless Networks

Edge Cloud Computing and Federated–Split Learning in Internet of Things

Combining Federated Learning and Edge Computing Toward Ubiquitous Intelligence in 6G Network: Challenges, Recent Advances, and Future Directions

Hierarchical Federated Edge Learning with Adaptive Clustering in Internet of Things

Edge-assisted U-Shaped Split Federated Learning with Privacy-preserving for Internet of Things

Survey: federated learning data security and privacy-preserving in edge-Internet of Things

Federated Split Learning for Distributed Intelligence with Resource-Constrained Devices

Federated Learning in Mobile Edge Computing: An Edge-Learning Perspective for Beyond 5G

Split Learning in 6G Edge Networks

Model-Heterogeneous Federated Learning for Internet of Things: Enabling Technologies and Future Directions

Semi-Federated Learning for Internet of Intelligence

Wireless Communications for Collaborative Federated Learning

Semi-Federated Learning for Collaborative Intelligence in Massive IoT Networks

Personalized Federated Learning for Intelligent IoT Applications: A Cloud-Edge based Framework

A Comprehensive Survey on Joint Resource Allocation Strategies in Federated Edge Learning

PPSFL: Privacy-Preserving Split Federated Learning for heterogeneous data in edge-based Internet of Things

Federated Learning for Internet of Things: Recent Advances, Taxonomy, and Open Challenges

Evaluation and Optimization of Distributed Machine Learning Techniques for Internet of Things