Searcharxiv⌕ Search

arXiv subjects

Renaud Sirdey

Publications and source records attributed to Renaud Sirdey.

9 recordsLinked to original sources
arXiv2026

Robust Federated Learning via Byzantine Filtering over Encrypted Updates

Federated Learning (FL) aims to train a collaborative model while preserving data privacy. However, the distributed nature of this approach still raises privacy and security issues, such as the exposure of sensitive data due to inference attacks and the influence of Byzantine behaviors on the trained model. In particular, achieving both secure aggregation and Byzantine resilience remains challenging, as existing solutions often address these aspects independently. In this work, we propose to address these challenges through a novel approach that combines homomorphic encryption for privacy-preserving aggregation with property-inference-inspired meta-classifiers for Byzantine filtering. First, following the property-inference attacks blueprint, we train a set of filtering meta-classifiers on labeled shadow updates, reproducing a diverse ensemble of Byzantine misbehaviors in FL, including backdoor, gradient-inversion, label-flipping and shuffling attacks. The outputs of these meta-classifiers are then used to cancel the Byzantine encrypted updates by reweighting. Second, we propose an automated method for selecting the optimal kernel and the dimensionality hyperparameters with respect to homomorphic inference, aggregation constraints and efficiency over the CKKS cryptosystem. Finally, we demonstrate through extensive experiments the effectiveness of our approach against Byzantine participants on the FEMNIST, CIFAR10, GTSRB, and acsincome benchmarks. More precisely, our SVM filtering achieves accuracies between $90$% and $94$% for identifying Byzantine updates at the cost of marginal losses in model utility and encrypted inference runtimes ranging from $6$ to $24$ seconds and from $9$ to $26$ seconds for an overall aggregation.

cs.LG↗
arXiv2023

SABLE: Secure And Byzantine robust LEarning

Due to the widespread availability of data, machine learning (ML) algorithms are increasingly being implemented in distributed topologies, wherein various nodes collaborate to train ML models via the coordination of a central server. However, distributed learning approaches face significant vulnerabilities, primarily stemming from two potential threats. Firstly, the presence of Byzantine nodes poses a risk of corrupting the learning process by transmitting inaccurate information to the server. Secondly, a curious server may compromise the privacy of individual nodes, sometimes reconstructing the entirety of the nodes' data. Homomorphic encryption (HE) has emerged as a leading security measure to preserve privacy in distributed learning under non-Byzantine scenarios. However, the extensive computational demands of HE, particularly for high-dimensional ML models, have deterred attempts to design purely homomorphic operators for non-linear robust aggregators. This paper introduces SABLE, the first homomorphic and Byzantine robust distributed learning algorithm. SABLE leverages HTS, a novel and efficient homomorphic operator implementing the prominent coordinate-wise trimmed mean robust aggregator. Designing HTS enables us to implement HMED, a novel homomorphic median aggregator. Extensive experiments on standard ML tasks demonstrate that SABLE achieves practical execution times while maintaining an ML accuracy comparable to its non-private counterpart.

cs.LG↗
arXiv2023

When approximate design for fast homomorphic computation provides differential privacy guarantees

While machine learning has become pervasive in as diversified fields as industry, healthcare, social networks, privacy concerns regarding the training data have gained a critical importance. In settings where several parties wish to collaboratively train a common model without jeopardizing their sensitive data, the need for a private training protocol is particularly stringent and implies to protect the data against both the model's end-users and the actors of the training phase. Differential privacy (DP) and cryptographic primitives are complementary popular countermeasures against privacy attacks. Among these cryptographic primitives, fully homomorphic encryption (FHE) offers ciphertext malleability at the cost of time-consuming operations in the homomorphic domain. In this paper, we design SHIELD, a probabilistic approximation algorithm for the argmax operator which is both fast when homomorphically executed and whose inaccuracy is used as a feature to ensure DP guarantees. Even if SHIELD could have other applications, we here focus on one setting and seamlessly integrate it in the SPEED collaborative training framework from "SPEED: Secure, PrivatE, and Efficient Deep learning" (Grivet Sébert et al., 2021) to improve its computational efficiency. After thoroughly describing the FHE implementation of our algorithm and its DP analysis, we present experimental results. To the best of our knowledge, it is the first work in which relaxing the accuracy of an homomorphic calculation is constructively usable as a degree of freedom to achieve better FHE performances.

cs.CR↗
arXiv2023

Homomorphic Sortition -- Secret Leader Election for PoS Blockchains

In a single secret leader election protocol (SSLE), one of the system participants is chosen and, unless it decides to reveal itself, no other participant can identify it. SSLE has a great potential in protecting blockchain consensus protocols against denial of service (DoS) attacks. However, all existing solutions either make strong synchrony assumptions or have expiring registration, meaning that they require elected processes to re-register themselves before they can be re-elected again. This, in turn, prohibits the use of these SSLE protocols to elect leaders in partially-synchronous consensus protocols as there may be long periods of network instability when no new blocks are decided and, thus, no new registrations (or re-registrations) are possible. In this paper, we propose Homomorphic Sortition -- the first asynchronous SSLE protocol with non-expiring registration, making it the first solution compatible with partially-synchronous leader-based consensus protocols. Homomorphic Sortition relies on Threshold Fully Homomorphic Encryption (ThFHE) and is tailored to proof-of-stake (PoS) blockchains, with several important optimizations with respect to prior proposals. In particular, unlike most existing SSLE protocols, it works with arbitrary stake distributions and does not require a user with multiple coins to be registered multiple times. Our protocol is highly parallelizable and can be run completely off-chain after setup. Some blockchains require a sequence of rounds to have non-repeating leaders. We define a generalization of SSLE, called Secret Leader Permutation (SLP) in which the application can choose how many non-repeating leaders should be output in a sequence of rounds and we show how Homomorphic Sortition also solves this problem.

cs.CR↗
arXiv2022

Protecting Data from all Parties: Combining FHE and DP in Federated Learning

This paper tackles the problem of ensuring training data privacy in a federated learning context. Relying on Homomorphic Encryption (HE) and Differential Privacy (DP), we propose a framework addressing threats on the privacy of the training data. Notably, the proposed framework ensures the privacy of the training data from all actors of the learning process, namely the data owners and the aggregating server. More precisely, while HE blinds a semi-honest server during the learning protocol, DP protects the data from semi-honest clients participating in the training process as well as end-users with black-box or white-box access to the trained model. In order to achieve this, we provide new theoretical and practical results to allow these techniques to be rigorously combined. In particular, by means of a novel stochastic quantisation operator, we prove DP guarantees in a context where the noise is quantised and bounded due to the use of HE. The paper is concluded by experiments which show the practicality of the entire framework in terms of both model quality (impacted by DP) and computational overhead (impacted by HE).

cs.CR↗
arXiv2021

RandSolomon: Optimally Resilient Random Number Generator with Deterministic Termination

Multi-party random number generation is a key building-block in many practical protocols. While straightforward to solve when all parties are trusted to behave correctly, the problem becomes much more difficult in the presence of faults. In this context, this paper presents RandSolomon, a protocol that allows a network of N processes to produce an unpredictable common random number among the non-faulty of them. We provide optimal resilience for partially-synchronous systems where less than a third of the participants might behave arbitrarily and, contrary to many solutions, we do not require at any point faulty-processes to be responsive.

cs.DC↗
arXiv2021

SPEED: Secure, PrivatE, and Efficient Deep learning

We introduce a deep learning framework able to deal with strong privacy constraints. Based on collaborative learning, differential privacy and homomorphic encryption, the proposed approach advances state-of-the-art of private deep learning against a wider range of threats, in particular the honest-but-curious server assumption. We address threats from both the aggregation server, the global model and potentially colluding data holders. Building upon distributed differential privacy and a homomorphic argmax operator, our method is specifically designed to maintain low communication loads and efficiency. The proposed method is supported by carefully crafted theoretical results. We provide differential privacy guarantees from the point of view of any entity having access to the final model, including colluding data holders, as a function of the ratio of data holders who kept their noise secret. This makes our method practical to real-life scenarios where data holders do not trust any third party to process their datasets nor the other data holders. Crucially the computational burden of the approach is maintained reasonable, and, to the best of our knowledge, our framework is the first one to be efficient enough to investigate deep learning applications while addressing such a large scope of threats. To assess the practical usability of our framework, experiments have been carried out on image datasets in a classification context. We present numerical results that show that the learning procedure is both accurate and private.

cs.CR↗
arXiv2019

Revisiting old combinatorial beasts in the quantum age: quantum annealing versus maximal matching

This paper experimentally investigates the behavior of analog quantum computers such as commercialized by D-Wave when confronted to instances of the maximum cardinality matching problem specifically designed to be hard to solve by means of simulated annealing. We benchmark a D-Wave "Washington" (2X) with 1098 operational qubits on various sizes of such instances and observe that for all but the most trivially small of these it fails to obtain an optimal solution. Thus, our results suggests that quantum annealing, at least as implemented in a D-Wave device, falls in the same pitfalls as simulated annealing and therefore suggest that there exist polynomial-time problems that such a machine cannot solve efficiently to optimality.

quant-ph↗