Details

Autor(en) / Beteiligte

• Wang, Haotian
• Zheng, Dongyi
• Liu, Fang
• Xiao, Nong

Titel

BlockFAD: A Federated Learning Based I/O Anomaly Detection Method for Blockchain Domain Name System

Ist Teil von

• Advances in Artificial Intelligence and Security, p.668-679

Ort / Verlag

Cham: Springer International Publishing

Link zum Volltext

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• Anomaly detection in storage systems of blockchain domain name service is a challenging problem due to the high dimensional sequential data involved, lack of labels, and privacy data protection issues in distributed storage. To find anomalies in the massively distributed domain name servers, we present BlockFAD, a federated learning based I/O anomaly detection method for blockchain DNS. The distributed model aggregation scheme based on the blockchain ensures the security of the model sharing and storage. And we use the two-channel Transformer encoder to detect anomalies in DNS I/O. Experiments on datasets such as server anomaly detection demonstrate the effectiveness and rationality of the method.