The article discusses the use of blockchain technology to address data privacy and security issues in collaborative machine learning over Internet of Things (IoT) devices. Traditional centralized machine learning models face challenges such as data privacy, security, and scalability. Federated learning is a technique that enables decentralized model training on distributed data sources while preserving data privacy and security. However, it faces challenges like data poisoning attacks, model inversion attacks, etc. Blockchain technology can mitigate these risks by providing a secure and transparent platform to store and manage the global model and model updates from each participating device. The article introduces an IoT solution that combines blockchain technology and federated learning for collaborative machine learning.
Publication date: 27 Nov 2023
Project Page: Not Provided
Paper: https://arxiv.org/pdf/2311.14136
