Model poisoning attack in differential privacy-based federated learning

Yang, Ming–Hsuan; Cheng, Hang; Chen, Fei; Liu, Ximeng; Wang, Meiqing; Li, Xibin

doi:10.1016/j.ins.2023.02.025

Cited by 23 publications

(8 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…More generally, the fact that the official network is, by design, produced to assist the secret network in extracting information might allow the framework to find ways around defence mechanisms that have been proven successful against similar attacks, such as model inversion ones. On the other hand, differential privacy [18], together with strategies to decouple data from model training [14], should prove successful in protecting against it.…”

Section: Discussionmentioning

confidence: 99%

Neural Network Information Leakage Through Hidden Learning

Cunha

Natale

Viennot³

2023

Communications in Computer and Information Science

View full text Add to dashboard Cite

We investigate the problem of making an artificial neural network perform hidden computations whose result can be easily retrieved from the network's output. In particular, we consider the following scenario. A user is provided with a neural network for a classification task by a third party. The user's input to the network contains sensitive information and the third party can only observe the output of the network. In this work, we provide a simple and efficient training procedure, which we call hidden learning, that produces two networks: (i) one that solves the original classification task with performance near to state of the art; (ii) a second one that takes as input the output of the first, retrieving sensitive information to solve a second classification task with good accuracy. Our result might expose important issues from an information security point of view, as well as for the use of artificial neural networks in sensible applications.

show abstract

Section: Discussionmentioning

confidence: 99%

Neural Network Information Leakage Through Hidden Learning

Cunha

Natale

Viennot³

2023

Communications in Computer and Information Science

View full text Add to dashboard Cite

show abstract

“…In gradient ascent attacks, the attacker updates the model in the direction that maximizes the loss. The model shuffling attack aims at shuffling the model parameters without notably changing the loss [28].…”

Section: B Poisoning Attacks In Federated Learningmentioning

confidence: 99%

Poisoning Attacks on Federated Learning for Autonomous Driving

Garg,

Jönsson,

Kalander

et al. 2024

Linköping Electronic Conference Proceedings

View full text Add to dashboard Cite

Federated Learning (FL) is a decentralized learning paradigm, enabling parties to collaboratively train models while keeping their data confidential. Within autonomous driving, it brings the potential of reducing data storage costs, reducing bandwidth requirements, and to accelerate the learning. FL is, however, susceptible to poisoning attacks. In this paper, we introduce two novel poisoning attacks on FL tailored to regression tasks within autonomous driving: FLStealth and Off-Track Attack (OTA). FLStealth, an untargeted attack, aims at providing model updates that deteriorate the global model performance while appearing benign. OTA, on the other hand, is a targeted attack with the objective to change the global model's behavior when exposed to a certain trigger. We demonstrate the effectiveness of our attacks by conducting comprehensive experiments pertaining to the task of vehicle trajectory prediction. In particular, we show that, among five different untargeted attacks, FLStealth is the most successful at bypassing the considered defenses employed by the server.For OTA, we demonstrate the inability of common defense strategies to mitigate the attack, highlighting the critical need for new defensive mechanisms against targeted attacks within FL for autonomous driving.

show abstract

“…Generally, most of the federated learning studies focus on looking for the best compromise between privacy and utility [ 63 ]. However, another crucial issue that must be considered is the malicious security threats encountered by federated learning [ 64 ].…”

Section: Software-based Solutionsmentioning

confidence: 99%

Genomic privacy preservation in genome-wide association studies: taxonomy, limitations, challenges, and vision

Aherrahrou,

Tairi,

Aherrahrou

2024

Briefings in Bioinformatics

View full text Add to dashboard Cite

Genome-wide association studies (GWAS) serve as a crucial tool for identifying genetic factors associated with specific traits. However, ethical constraints prevent the direct exchange of genetic information, prompting the need for privacy preservation solutions. To address these issues, earlier works are based on cryptographic mechanisms such as homomorphic encryption, secure multi-party computing, and differential privacy. Very recently, federated learning has emerged as a promising solution for enabling secure and collaborative GWAS computations. This work provides an extensive overview of existing methods for GWAS privacy preserving, with the main focus on collaborative and distributed approaches. This survey provides a comprehensive analysis of the challenges faced by existing methods, their limitations, and insights into designing efficient solutions.

show abstract

Model poisoning attack in differential privacy-based federated learning

Cited by 23 publications

References 14 publications

Neural Network Information Leakage Through Hidden Learning

Neural Network Information Leakage Through Hidden Learning

Poisoning Attacks on Federated Learning for Autonomous Driving

Genomic privacy preservation in genome-wide association studies: taxonomy, limitations, challenges, and vision

Contact Info

Product

Resources

About