Название: Split Federated Learning for Secure IoT Applications: Concepts, frameworks, applications and case studies
Автор: Gururaj Harinahalli Lokesh, Geetabai S. Hukkeri, N.Z. Jhanjhi, Hong Lin
Издательство: The Institution of Engineering and Technology
Год: 2024
Страниц: 285
Язык: английский
Формат: pdf (true)
Размер: 10.1 MB

New approaches in Federated Learning and Split Learning have the potential to significantly improve ubiquitous intelligence in Internet of Things (IoT) applications. In Split Federated Learning, the Machine Learning model is divided into smaller network segments, with each segment trained independently on a server using distributed local client data.

The split learning method mitigates two fundamental drawbacks of Federated Learning: affordability, and privacy and security. When running Machine Learning computation on devices with limited resources, assigning only a portion of the network to train at the client-side minimizes the processing burden, compared to running a complete network as in Federated Learning. In addition, neither client nor server has full access to the other, which is more secure.

This book reviews cutting edge technologies and advanced research in Split Federated Learning. Coverage includes approaches to realizing and evaluating the effectiveness and advantages of Federated Learning and Splitfed learning, the role of this technology in advancing and securing IoTs, advanced research on emerging AI models for preserving the privacy of the data owned by the clients, and the analysis and development of AI mechanisms in IoT architectures and applications. The use of Split Federated Learning in natural language processing (NLP), recommendation systems, healthcare systems, emotion detection, smart agriculture, smart transportation and smart cities is discussed.

The authors of Chapter 1 elaborate on the fundamentals of Federated Learning (FL) and Split Learning which are the two common approaches to distributed Machine Learning. This chapter introduces Splitfed Learning, a novel approach that combines the two abovementioned approaches while eliminating the resulting drawbacks, as well as a refined structural configuration including distinct security and PixelDP to improve data privacy and durability.

In Chapter 2, the authors have discussed an overview of the cutting-edge technologies for integrating these FL approaches into an edge computing-based IoT setting. Furthermore, it addresses some existing challenges and also discusses future research directions to spark further exploration within the academic community.

The authors of Chapters 3 and 4 provide an overview of blockchain-driven Splitfed Learning for data protection in IoT setting and Splitfed Learning methods for natural language processing.

In Chapters 5 and 6, the authors are concerned with the role of Splitfed Learning in recommendation systems. Delving into the applications of RIS within the context of 6G, various configurations of RIS-assisted wireless systems are scrutinized, covering diverse scenarios, system and fading models, as well as performance metrics and objectives, in a comprehensive and methodical manner.
...
In Chapters 11 and 12, the authors have discussed the case study of Splitfed Learning for smart grids and smart agriculture. Splitfed Learning enables immediate improvements through localized model updates, promotes data privacy by retaining control over sensitive information within individual components and enhances the overall efficiency and resilience by leveraging diverse data sources.

Split Federated Learning for Secure IoT Applications: Concepts, frameworks, applications and case studies offers useful insights to the latest developments in the field for researchers, engineers and scientists in academia and industry, who are working in computing, AI, Data Science and cybersecurity with a focus on Federated Learning, Machine Learning and Deep Learning.




ОТСУТСТВУЕТ ССЫЛКА/ НЕ РАБОЧАЯ ССЫЛКА ЕСТЬ РЕШЕНИЕ, ПИШИМ СЮДА!