Robust Privacy-Preserving Gossip Averaging

Pilet, Amaury Bouchra; Frey, Davide; Taı̈ani, François

doi:10.1007/978-3-030-34992-9_4

Cited by 5 publications

(3 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One notable approach is proposed in [25], where each node decomposes its state into two sub-states to guarantee privacy in average consensus. The additive-noise strategy presented in [26] exchanges noise for several rounds before exchanging data to ensure robustness, but it is only partially immune to attacks. Another interesting protection mechanism is digital watermarking [27], which can be used to protect signals from unauthorized access, copying, and distribution.…”

Section: A State Of the Artmentioning

confidence: 99%

Exploiting the Synchronization of Nonlinear Dynamics to Secure Distributed Consensus

Fioravanti

Bonagura

Oliva

et al. 2023

IEEE Open J. Control. Syst.

View full text Add to dashboard Cite

Distributed cooperative multi-agent operations, which are emerging as effective solutions in countless application domains, are prone to eavesdropping by malicious entities due to their exposure on the network. Moreover, in several cases, agents are reluctant to disclose their initial conditions (even to legitimate neighbors) due to their sensitivity to private data. Providing security guarantees against external readings by malicious entities and the privacy of exchanged data while allowing agents to reach an agreement on some shared variables is an essential feature to foster the adoption of distributed protocols. In this article, we propose to implement a secure and privacy-preserving consensus strategy that exploits, for this purpose, the performance of synchronization of nonlinear continuous-time dynamical systems. This is achieved by splitting the initial conditions into two information fragments, one of which is subject to nonlinear manipulation. In this way, the information being exchanged in the network will always be subject to the influence of nonlinear dynamics. However, by exploiting the ability of such dynamics to synchronize, the combination of the two fragments still converges to a weighted average of each node's actual initial conditions. Furthermore, due to the dependence of the hidden dynamics on a coordinate transformation known only to the legitimate nodes, message security is ensured even once consensus is reached; our approach relies on the assumption that a secure communication channel is available during an initialization phase. The article is complemented by a simulation campaign aimed at numerically demonstrating the effectiveness of the proposed approach.

show abstract

Section: A State Of the Artmentioning

confidence: 99%

Exploiting the Synchronization of Nonlinear Dynamics to Secure Distributed Consensus

Fioravanti

Bonagura

Oliva

et al. 2023

IEEE Open J. Control. Syst.

View full text Add to dashboard Cite

show abstract

“…The first averages all the elements of a set of vectors, the second averages predefined coordinates (second parameter) of the vectors from a set (first parameter). In a decentralized setup, these functions can be implemented with a gossip-based averaging protocol [JMB05], or with a private one [BFT19] for improved privacy protection. It is also possible to implement these primitives on a central server, to which peers send their models for averaging, in a federated, rather than decentralized, setup.…”

Section: Ii3b Algorithmmentioning

confidence: 99%

“…Also, peers will never get to see any part of another peer's model with a centralized implementation. When implementing the method using gossip averaging, a private gossip averaging protocol, like the one proposed in [BFT19], can be used.…”

Section: Ii5 Privacymentioning

confidence: 99%

Simple, Efficient and Convenient Decentralized Multi-task Learning for Neural Networks

Pilet

Frey

Taı̈ani

2021

Advances in Intelligent Data Analysis XIX

Self Cite

View full text Add to dashboard Cite

Machine learning requires large amounts of data, which is increasingly distributed over many systems (user devices, independent storage systems). Unfortunately aggregating this data in one site for learning is not always practical, either because of network costs or privacy concerns. Decentralized machine learning holds the potential to address these concerns, but unfortunately, most approaches proposed so far for distributed learning with neural network are mono-task, and do not transfer easily to multi-tasks problems, for which users seek to solve related but distinct learning tasks and the few existing multi-task approaches have serious limitations. In this paper, we propose a novel learning method for neural networks that is decentralized, multi-task, and keeps users' data local. Our approach works with different learning algorithms, on various types of neural networks. We formally analyze the convergence of our method, and we evaluate its efficiency in different situations on various kind of neural networks, with different learning algorithms, thus demonstrating its benefits in terms of learning quality and convergence. I IntroductionA critical requirement for machine learning is training data. In some cases, a great amount of data is available from different (potential) users of the system, but simply aggregating this data and using it for training is not always practical. The data might for instance be large and extremely distributed and collecting it may incur significant communication cost. Users may also be unwilling to share their data due to its potential sensitive nature, as is the case with private conversations, browsing histories, or health-related data [Che+17].To address these issues, several works have proposed to share model-related information (such as gradients or model coefficients) rather than raw data [Rec+11; Kon+16; Bre+17]. These approaches are however typically mono-task, in the sense that all users are assumed to be solving the same ML task. Unfortunately, in a distributed setting, the problems that users want to solve may not be perfectly identical. Let us consider the example of speech recognition being performed on mobile device. At the user level, each has a different voice, and so the different devices do not perform exactly the same task. At the level of a country or region, language variants also impose variants between tasks. For example users from Quebec can clearly be separated from those from France. A decentralized or federated learning platform should therefore accommodate for both types of differences between tasks: at the level of single users and at the level of groups of users.Multitask learning has been widely studied in a centralized setting [Rud17]. Some distributed solutions exist, but they are typically limited to either linear [OHJ12] or convex [Smi+16; Bel+18; ZBT19] optimization problems. As a result, they are typically not applicable to neural networks, in spite of their high versatility and general success in solving a broad range of machine-learning 1/39 D...

show abstract

Foiling Sybils with HAPS in Permissionless Systems: An Address-based Peer Sampling Service

Pilet

Frey

Taı̈ani

2020

2020 IEEE Symposium on Computers and Communications (ISCC)

Self Cite

View full text Add to dashboard Cite

show abstract

Robust Privacy-Preserving Gossip Averaging

Cited by 5 publications

References 34 publications

Exploiting the Synchronization of Nonlinear Dynamics to Secure Distributed Consensus

Exploiting the Synchronization of Nonlinear Dynamics to Secure Distributed Consensus

Simple, Efficient and Convenient Decentralized Multi-task Learning for Neural Networks

Foiling Sybils with HAPS in Permissionless Systems: An Address-based Peer Sampling Service

Contact Info

Product

Resources

About