UAI 2019poster329 citations
Fall of Empires: Breaking Byzantine-tolerant SGD by Inner Product Manipulation
Cong Xie, Oluwasanmi Koyejo, Indranil Gupta
Abstract
Recently, new defense techniques have been developed to tolerate Byzantine failures for distributed machine learning. The Byzantine model captures workers that behave arbitrarily, including malicious and compromised workers. In this paper, we break two prevailing Byzantine-tolerant techniques. Specifically we show that two robust aggregation methods for synchronous SGD–namely, coordinate-wise median and Krum–can be broken using new attack strategies based on inner product manipulation. We prove our results theoretically, as well as show empirical validation.
BibTeX
@InProceedings{pmlr-v115-xie20a,
title = {Fall of Empires: Breaking Byzantine-tolerant SGD by Inner Product Manipulation},
author = {Xie, Cong and Koyejo, Oluwasanmi and Gupta, Indranil},
booktitle = {Proceedings of The 35th Uncertainty in Artificial Intelligence Conference},
pages = {261--270},
year = {2020},
editor = {Adams, Ryan P. and Gogate, Vibhav},
volume = {115},
series = {Proceedings of Machine Learning Research},
month = {22--25 Jul},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v115/xie20a/xie20a.pdf},
url = {https://proceedings.mlr.press/v115/xie20a.html},
abstract = {Recently, new defense techniques have been developed to tolerate Byzantine failures for distributed machine learning. The Byzantine model captures workers that behave arbitrarily, including malicious and compromised workers. In this paper, we break two prevailing Byzantine-tolerant techniques. Specifically we show that two robust aggregation methods for synchronous SGD–namely, coordinate-wise median and Krum–can be broken using new attack strategies based on inner product manipulation. We prove our results theoretically, as well as show empirical validation. }
}