A Verifiable Federated Learning Mechanism for Digital Museum Collection Based on Homomorphic Proxy Re-Authentication Technology

Despite the opportunities presented by information technology and machine learning for the digitization of natural heritage, challenges such as insufficient data sets and inadequate security remain significant obstacles. To address these challenges, this article proposes a cross-regional intrusion detection federated learning scheme based on homomorphic encryption-based re-authentication technology. By integrating homomorphic encryption-based re-authentication, secret sharing, and key negotiation technologies, the scheme achieves an efficient and verifiable private gradient aggregation solution. Furthermore, the scheme introduces intermediary proxy nodes, thwarting collusive attacks between servers and participants and significantly enhancing the overall security of the system. Through these mechanisms, various issues related to privacy protection, data leakage, data forgery, and access control when multiple museums share collection data are effectively addressed. By implementing the designed scheme and amalgamating data from the National Museum of Natural History and the National Museum of Animals in China, a virtual insect exhibition hall has been developed. This endeavor serves to facilitate the creation of an information model for insects within the metaverse. Furthermore, due to the inherent characteristics of the scheme, not only is data security ensured, but it also fosters collaboration between museums, advancing the digital preservation and sharing of cultural heritage.