DarkneTZ

Mo, Fan; Shamsabadi, Ali Shahin; Katevas, Kleomenis; Demetriou, Soteris; Leontiadis, Ilias; Cavallaro, Andrea; Haddadi, Hamed

doi:10.1145/3386901.3388946

Cited by 91 publications

(34 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…TEE provides a promising hardware-based alternative for achieving secure computation with a CPUlevel performance, and thus bridges the gap between academic research and industrial adoption of secure computation. Trusted hardware provides a powerful abstraction for building secure systems and can be applied to many potential applications, such as machine learning algorithms [81,97,98,102,106,142], cloud computing [15,33,94,122,126,141], blockchain [18,22,87,159], network traffic analysis [47], scientific computation [130], data analytics [41,75,118,164,165], and privacy-preserving COVID-19 contact tracing [73].…”

Section: Our Contributionsmentioning

confidence: 99%

“…However, due to the partial in-enclave training, SecureFL results in 1.6%-23.6% overhead compared with a pure CPU-based training. Aiming to limit privacy leakages in FL, Mo et al presented a privacy-preserving federated learning (PPFL) framework [98] for mobile systems using TEE. PPFL adopted TEE to train each layer of the model until its convergence by leveraging the greedy layerwise.…”

Section: Research Statusmentioning

confidence: 99%

See 1 more Smart Citation

A Survey of Secure Computation Using Trusted Execution Environments

Li¹,

Yang²,

Xiang³

et al. 2023

Preprint

View full text Add to dashboard Cite

As an essential technology underpinning trusted computing, the trusted execution environment (TEE) allows one to launch computation tasks on both on-and off-premises data while assuring confidentiality and integrity. This article provides a systematic review and comparison of TEE-based secure computation protocols. We first propose a taxonomy that classifies secure computation protocols into three major categories, namely secure outsourced computation, secure distributed computation and secure multi-party computation. To enable a fair comparison of these protocols, we also present comprehensive assessment criteria with respect to four aspects: setting, methodology, security and performance. Based on these criteria, we review, discuss and compare the state-of-the-art TEE-based secure computation protocols for both general-purpose computation functions and special-purpose ones, such as privacy-preserving machine learning and encrypted database queries. To the best of our knowledge, this article is the first survey to review TEE-based secure computation protocols and the comprehensive comparison can serve as a guideline for selecting suitable protocols for deployment in practice. Finally, we also discuss several future research directions and challenges.CCS Concepts: • Security and privacy → Tamper-proof and tamper-resistant designs; Trust frameworks; Privacy-preserving protocols; Management and querying of encrypted data.

show abstract

Section: Our Contributionsmentioning

confidence: 99%

Section: Research Statusmentioning

confidence: 99%

A Survey of Secure Computation Using Trusted Execution Environments

Li¹,

Yang²,

Xiang³

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…On the other hand, defenses against inversion attacks have seen a surge since the introduction of FL. DarkneTZ [31], PPFL [30] and GradSec [29] mitigate these attacks with the use of a TEE. By protecting sensitive parameters, activations and gradients inside the enclave memory, the risk of gradient leakage is alleviated and the white-box setting is effectively limited, thus weakening the threat.…”

Section: Related Workmentioning

confidence: 99%

Pelta

Queyrut

Bromberg

Schiavoni

2022

Proceedings of the 3rd International Workshop on Distributed Machine Learning

View full text Add to dashboard Cite

The main premise of federated learning is that machine learning model updates are computed locally, in particular to preserve user data privacy, as those never leave the perimeter of their device. This mechanism supposes the general model, once aggregated, to be broadcast to collaborating and non malicious nodes. However, without proper defenses, compromised clients can easily probe the model inside their local memory in search of adversarial examples. For instance, considering image-based applications, adversarial examples consist of imperceptibly perturbed images (to the human eye) misclassified by the local model, which can be later presented to a victim node's counterpart model to replicate the attack. To mitigate such malicious probing, we introduce Pelta, a novel shielding mechanism leveraging trusted hardware. By harnessing the capabilities of Trusted Execution Environments (TEEs), Pelta masks part of the back-propagation chain rule, otherwise typically exploited by attackers for the design of malicious samples. We evaluate Pelta on a state of the art ensemble model and demonstrate its effectiveness against the Self Attention Gradient adversarial Attack.

show abstract

“…In addition to data privacy, model privacy also raises attention (Mo et al, 2020), which means foundation models can be black-box models (Guidotti et al, 2018;Ljung, 2001). Little work paid attention to finetuning or optimizing in this field, but most related work focused on attacks (Yang et al, 2023b;c).…”

Section: Related Workmentioning

confidence: 99%