اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدLayer-wise Characterization of Latent Information Leakage in Federated Learning
138
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Training deep neural networks via federated learning allows clients to share, instead of the original data, only the model trained on their data. Prior work has demonstrated that in practice a clients private information, unrelated to the main learning task, can be discovered from the models gradients, which compromises the promised privacy protection. However, there is still no formal approach for quantifying the leakage of private information via the shared updated model or gradients. In this work, we analyze property inference attacks and define two metrics based on (i) an adaptation of the empirical $mathcal{V}$-information, and (ii) a sensitivity analysis using Jacobian matrices allowing us to measure changes in the gradients with respect to latent information. We show the applicability of our proposed metrics in localizing private latent information in a layer-wise manner and in two settings where (i) we have or (ii) we do not have knowledge of the attackers capabilities. We evaluate the proposed metrics for quantifying information leakage on three real-world datasets using three benchmark models.
قيم البحث
اقرأ أيضاً
Federated learning enables mutually distrusting participants to collaboratively learn a distributed machine learning model without revealing anything but the models output. Generic federated learning has been studied extensively, and several learning
protocols, as well as open-source frameworks, have been developed. Yet, their over pursuit of computing efficiency and fast implementation might diminish the security and privacy guarantees of participants training data, about which little is known thus far. In this paper, we consider an honest-but-curious adversary who participants in training a distributed ML model, does not deviate from the defined learning protocol, but attempts to infer private training data from the legitimately received information. In this setting, we design and implement two practical attacks, reverse sum attack and reverse multiplication attack, neither of which will affect the accuracy of the learned model. By empirically studying the privacy leakage of two learning protocols, we show that our attacks are (1) effective - the adversary successfully steal the private training data, even when the intermediate outputs are encrypted to protect data privacy; (2) evasive - the adversarys malicious behavior does not deviate from the protocol specification and deteriorate any accuracy of the target model; and (3) easy - the adversary needs little prior knowledge about the data distribution of the target participant. We also experimentally show that the leaked information is as effective as the raw training data through training an alternative classifier on the leaked information. We further discuss potential countermeasures and their challenges, which we hope may lead to several promising research directions.
In the federated learning system, parameter gradients are shared among participants and the central modulator, while the original data never leave their protected source domain. However, the gradient itself might carry enough information for precise
inference of the original data. By reporting their parameter gradients to the central server, client datasets are exposed to inference attacks from adversaries. In this paper, we propose a quantitative metric based on mutual information for clients to evaluate the potential risk of information leakage in their gradients. Mutual information has received increasing attention in the machine learning and data mining community over the past few years. However, existing mutual information estimation methods cannot handle high-dimensional variables. In this paper, we propose a novel method to approximate the mutual information between the high-dimensional gradients and batched input data. Experimental results show that the proposed metric reliably reflect the extent of information leakage in federated learning. In addition, using the proposed metric, we investigate the influential factors of risk level. It is proven that, the risk of information leakage is related to the status of the task model, as well as the inherent data distribution.
A common goal in the areas of secure information flow and privacy is to build effective defenses against unwanted leakage of information. To this end, one must be able to reason about potential attacks and their interplay with possible defenses. In t
his paper, we propose a game-theoretic framework to formalize strategies of attacker and defender in the context of information leakage, and provide a basis for developing optimal defense methods. A novelty of our games is that their utility is given by information leakage, which in some cases may behave in a non-linear way. This causes a significant deviation from classic game theory, in which utility functions are linear with respect to players strategies. Hence, a key contribution of this paper is the establishment of the foundations of information leakage games. We consider two kinds of games, depending on the notion of leakage considered. The first kind, the QIF-games, is tailored for the theory of quantitative information flow (QIF). The second one, the DP-games, corresponds to differential privacy (DP).
Machine Learning models, extensively used for various multimedia applications, are offered to users as a blackbox service on the Cloud on a pay-per-query basis. Such blackbox models are commercially valuable to adversaries, making them vulnerable to
extraction attacks to reverse engineer the proprietary model thereby violating the model privacy and Intellectual Property. Here, the adversary first extracts the model architecture or hyperparameters through side channel leakage, followed by stealing the functionality of the target model by training the reconstructed architecture on a synthetic dataset. While the attacks proposed in literature are empirical, there is a need for a theoretical framework to measure the information leaked under such extraction attacks. To this extent, in this work, we propose a novel probabilistic framework, Airavata, to estimate the information leakage in such model extraction attacks. This framework captures the fact that extracting the exact target model is difficult due to experimental uncertainty while inferring model hyperparameters and stochastic nature of training to steal the target model functionality. Specifically, we use Bayesian Networks to capture uncertainty in estimating the target model under various extraction attacks based on the subjective notion of probability. We validate the proposed framework under different adversary assumptions commonly adopted in literature to reason about the attack efficacy. This provides a practical tool to infer actionable details about extracting blackbox models and help identify the best attack combination which maximises the knowledge extracted (or information leaked) from the target model.
The proposed pruning strategy offers merits over weight-based pruning techniques: (1) it avoids irregular memory access since representations and matrices can be squeezed into their smaller but dense counterparts, leading to greater speedup; (2) in a
manner of top-down pruning, the proposed method operates from a more global perspective based on training signals in the top layer, and prunes each layer by propagating the effect of global signals through layers, leading to better performances at the same sparsity level. Extensive experiments show that at the same sparsity level, the proposed strategy offers both greater speedup and higher performances than weight-based pruning methods (e.g., magnitude pruning, movement pruning).
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
