Highlight
Large-scale data acquisition is a critical issue for the development of industry-specific AI models, including deep learning and generative AI, trained with domain-specific adaptation. To address this issue, federated learning, a technique that enables the use of highly confidential data distributed across different organizations without compromising its confidentiality, has been proposed and developed.
This article describes how federated learning works and Hitachi's activities in this area. We also discuss its potential for use in generative AI, a field that has attracted considerable attention in recent years.
