paper-with-me

홈 › Papers

Data Poisoning against Differentially-Private Learners: Attacks and Defenses

2019-03-23 · Yuzhe Ma, Xiaojin Zhu, Justin Hsu

Data poisoning attacks aim to manipulate the model produced by a learning algorithm by adversarially modifying the training set. We consider differential privacy as a defensive measure against this type of attack. We show that such learners are resistant to data poisoning attacks when the adversary is only able to poison a small number of items. However, this protection degrades as the adversary poisons more data. To illustrate, we design attack algorithms targeting objective and output perturbation learners, two standard approaches to differentially-private machine learning. Experiments show that our methods are effective when the attacker is allowed to poison sufficiently many training items.

📄 PDF Abstract BibTeX arXiv:1903.09860

Code (0)

등록된 구현이 없습니다.

Tasks

Data Poisoning

Similar Papers 제목 키워드 기반

Auditing Differentially Private Machine Learning: How Private is Private SGD?

2020-06-13 · NeurIPS 2020 12 · Matthew Jagielski, Jonathan Ullman, Alina Oprea

We investigate whether Differentially Private SGD offers better privacy in practice than what is guaranteed by its state-of-the-art analysis. We do so via novel data poisoning attacks, which we show correspond to realist…

Art AnalysisBIG-bench Machine LearningData Poisoning

Differentially-Private Decision Trees and Provable Robustness to Data Poisoning

2023-05-24 · Daniël Vos, Jelle Vos, Tianyu Li, Zekeriya Erkin 외

Decision trees are interpretable models that are well-suited to non-linear learning problems. Much work has been done on extending decision tree learning algorithms with differential privacy, a system that guarantees the…

Data Poisoning

Strong Data Augmentation Sanitizes Poisoning and Backdoor Attacks Without an Accuracy Tradeoff

2020-11-18 · Eitan Borgnia, Valeriia Cherepanova, Liam Fowl, Amin Ghiasi 외

Data poisoning and backdoor attacks manipulate victim models by maliciously modifying training data. In light of this growing threat, a recent survey of industry professionals revealed heightened fear in the private sect…

Data AugmentationData Poisoning

DP-InstaHide: Provably Defusing Poisoning and Backdoor Attacks with Differentially Private Data Augmentations

2021-03-02 · Eitan Borgnia, Jonas Geiping, Valeriia Cherepanova, Liam Fowl 외

Data poisoning and backdoor attacks manipulate training data to induce security breaches in a victim model. These attacks can be provably deflected using differentially private (DP) training methods, although this comes …

Data Poisoning

Client Clustering Meets Knowledge Sharing: Enhancing Privacy and Robustness in Personalized Peer-to-Peer Learning

2025-06-25 · Mohammad Mahdi Maheri, Denys Herasymuk, Hamed Haddadi

The growing adoption of Artificial Intelligence (AI) in Internet of Things (IoT) ecosystems has intensified the need for personalized learning methods that can operate efficiently and privately across heterogeneous, reso…

Knowledge DistillationTransfer Learning