We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Computer Science > Machine Learning

Title: Byzantines can also Learn from History: Fall of Centered Clipping in Federated Learning

Abstract: The increasing popularity of the federated learning (FL) framework due to its success in a wide range of collaborative learning tasks also induces certain security concerns. Among many vulnerabilities, the risk of Byzantine attacks is of particular concern, which refers to the possibility of malicious clients participating in the learning process. Hence, a crucial objective in FL is to neutralize the potential impact of Byzantine attacks and to ensure that the final model is trustable. It has been observed that the higher the variance among the clients' models/updates, the more space there is for Byzantine attacks to be hidden. As a consequence, by utilizing momentum, and thus, reducing the variance, it is possible to weaken the strength of known Byzantine attacks. The centered clipping (CC) framework has further shown that the momentum term from the previous iteration, besides reducing the variance, can be used as a reference point to neutralize Byzantine attacks better. In this work, we first expose vulnerabilities of the CC framework, and introduce a novel attack strategy that can circumvent the defences of CC and other robust aggregators and reduce their test accuracy up to %33 on best-case scenarios in image classification tasks. Then, we propose a new robust and fast defence mechanism that is effective against the proposed and other existing Byzantine attacks.
Comments: IEEE Transactions on Information Forensics and Security 2023
Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (stat.ML)
DOI: 10.1109/TIFS.2023.3345171
Cite as: arXiv:2208.09894 [cs.LG]
  (or arXiv:2208.09894v3 [cs.LG] for this version)

Submission history

From: Kerem Özfatura [view email]
[v1] Sun, 21 Aug 2022 14:39:30 GMT (3379kb,D)
[v2] Sun, 30 Apr 2023 08:40:45 GMT (4889kb)
[v3] Mon, 1 Jan 2024 17:55:25 GMT (1583kb)

Link back to: arXiv, form interface, contact.