Paper 2024/506

A Decentralized Federated Learning using Reputation

Olive Chakraborty, CEA LIST
Aymen Boudguiga, CEA LIST

Nowadays Federated learning (FL) is established as one of the best techniques for collaborative machine learning. It allows a set of clients to train a common model without disclosing their sensitive and private dataset to a coordination server. The latter is in charge of the model aggregation. However, FL faces some problems, regarding the security of updates, integrity of computation and the availability of a server. In this paper, we combine some new ideas like clients’ reputation with techniques like secure aggregation using Homomorphic Encryption and verifiable secret sharing using Multi-Party Computation techniques to design a decentralized FL system that addresses the issues of incentives, security and availability amongst others. One of the original contributions of this work is the new leader election protocol which uses a secure shuffling and is based on a proof of reputation. Indeed, we propose to select an aggregator among the clients participating to the FL training using their reputations. That is, we estimate the reputation of each client at every FL iteration and then we select the next round aggregator from the set of clients with the best reputations. As such, we remove misbehaving clients (e.g., byzantines) from the list of clients eligible for the role of aggregation server.

Available format(s)
Publication info
Published elsewhere. ICISSP 2024
Decentralized federated learningreputationsecret-sharingfully homomorphic encryptionsecure shuffling
Contact author(s)
olive chakraborty @ cea fr
2024-04-01: approved
2024-03-29: received
See all versions
Short URL
Creative Commons Attribution


      author = {Olive Chakraborty and Aymen Boudguiga},
      title = {A Decentralized Federated Learning using Reputation},
      howpublished = {Cryptology ePrint Archive, Paper 2024/506},
      year = {2024},
      note = {\url{}},
      url = {}
Note: In order to protect the privacy of readers, does not use cookies or embedded third party content.