A Machine Learning Framework For Data Poisoning Attacks
DOI:
https://doi.org/10.46243/jst.2023.v8.i07.pp169-176Keywords:
Federated machine learning, Vulnerability, Arbitrary, Attack on federated machine learning(AT2FL), Gradients.Abstract
Federated models are built by collecting model changes from participants. To maintain the secrecy of the training data, the aggregator has no visibility into how these updates are made by design.. This paper aims to explore the vulnerability of federated machine learning, focusing on attacking a federated multitasking learning framework. The framework enables resource-constrained node devices, such as mobile phones and IOT devices, to learn a shared model while keeping the training However, the communication protocol among attackers may take advantage of various nodes to conduct data poisoning assaults, which has been shown to pose a serious danger to the majority of machine learning models. The paper formulates the problem of computing optimal poisoning attacks on federated multitask learning as a bi-level program that is adaptive to arbitrary choice of target nodes and source attacking nodes.The authors propose a novel systems-aware optimization method, Attack confederated Learning(AT2FL), which is efficiency to derive the implicit gradients for poisoned data and further compute optimal attack strategies in the federated machine learning.
References
Alfeld, S., Zhu, X., & Barford, P. (2016). Data Poisoning Attacks against Autoregressive Models. Proceedings of the AAAI Conference on Artificial Intelligence, 30(1). https://doi.org/10.1609/ aaai.v30i1.10237
Vitaly Shmatikov, Yiqing Hua, Deborah Estrin, Eugene Bagdasaryan, and Andreas Veit. federated learning backdoor techniques. 2018. arXiv preprint arXiv:1807.00459. https://doi. org/10.48550/arXiv.1807.00459
Barreno, M., Nelson, B., Joseph, A.D. et al. The security of machine learning. Mach Learn 81, 121–148(2010). https://doi.org/10.1007/ s10994-010-5188-5
Zhang S,Wang C and Zomaya C. Robustness Analysis and Enhancement of Deep Reinforcement Learning-Based Schedulers. IEEE Transactions on Parallel and Distributed Systems. 10.1109/ TPDS.2022.3218649. 34:1. (346-357 https:// ieeexplore.ieee.org/document/9937194/
Pavel Laskov, Blaine Nelson, and Batista Biggio. assaults with poison on support vector machines. 2012’s arXiv preprint 1206.6389 https://doi. org/10.48550/arXiv.1206.6389
Andrew Zisserman, Karen Simonyan, Ken Chatfield, and Andrea Vedaldi. The devil is once again in the details: a thorough examination of convolution networks. 2014 arXiv preprint 1405.3531. https://doi.org/10.48550/ arXiv.1405.353
Massimo Fornasier, Ignace Loris, and Ingrid Daubechies. Accelerated projected gradient approach for sparsity-constrained linear inverse problems. 2008;14(5-6):764–792;Journal of Fourier Analysis and Applications. https://doi. org/10.48550/arXiv.0706.4297
Mariana Raykova, Phillip Schoppmann, and Adrià Gascón. On datasets that have been vertically partitioned, secure linear regression. https://doi.org/10.48550/arXiv.0706.4297
S. Han, W. K. Ng, L. Wan and V. C. S. Lee, “Privacy-Preserving Gradient-Descent Methods,” in IEEE Transactions on Knowledge and Data Engineering, vol. 22, no. 6, pp. 884-899, June 2010, doi: 10.1109/TKDE.2009.153
Ling Huang, JD Tygar, Anthony D. Joseph, Blaine Nelson, and Benjamin I. P. Rubinstein. Machine learning that is hostile. Pages 43–58 in the book Proceedings of the Fourth ACM Workshop on Security and Artificial Intelligence. ACM, 2011. https://ieeexplore.ieee.org/ document/9887796/
