TDFL: Truth Discovery Based Byzantine Robust Federated Learning

• Published in: IEEE transactions on parallel and distributed systems Vol. 33; no. 12; pp. 1 - 14
• Main Authors: Xu, Chang, Jia, Yu, Zhu, Liehuang, Zhang, Chuan, Jin, Guoxie, Sharif, Kashif
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.12.2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Agglomeration; Collaborative work; Data models; Data privacy; Datasets; Environment models; Federated learning; Poisoning; poisoning attack; Robustness; Servers; Soft sensors; Training; truth discovery

Federated learning (FL) enables data owners to train a joint global model without sharing private data. However, it is vulnerable to Byzantine attackers that can launch poisoning attacks to destroy model training. Existing defense strategies rely on the additional datasets to train trustable server...