ML Anthology

SAMBA: A Generic Framework for Secure Federated Multi-Armed Bandits (Extended Abstract)

Radu Ciucanu, Pascal Lafourcade, Gaël Marcadet, Marta Soare
IJCAI 2023 pp. 6863-6867
doi:10.24963/IJCAI.2023/772 /ijcai/2023/ciucanu2023ijcai-samba/

Abstract

We tackle the problem of secure cumulative reward maximization in multi-armed bandits in a cross-silo federated learning setting. Under the orchestration of a central server, each data owner participating at the cumulative reward computation has the guarantee that its raw data is not seen by some other participant. We rely on cryptographic schemes and propose SAMBA, a generic framework for Secure federAted Multi-armed BAndits. We show that SAMBA returns the same cumulative reward as the non-secure versions of bandit algorithms, while satisfying formally proven security properties. We also show that the overhead due to cryptographic primitives is linear in the size of the input, which is confirmed by our implementation.

PDF IJCAI Semantic Scholar

Cite

Text

Ciucanu et al. "SAMBA: A Generic Framework for Secure Federated Multi-Armed Bandits (Extended Abstract)." International Joint Conference on Artificial Intelligence, 2023. doi:10.24963/IJCAI.2023/772

Markdown

[Ciucanu et al. "SAMBA: A Generic Framework for Secure Federated Multi-Armed Bandits (Extended Abstract)." International Joint Conference on Artificial Intelligence, 2023.](https://mlanthology.org/ijcai/2023/ciucanu2023ijcai-samba/) doi:10.24963/IJCAI.2023/772

BibTeX

@inproceedings{ciucanu2023ijcai-samba,
  title     = {{SAMBA: A Generic Framework for Secure Federated Multi-Armed Bandits (Extended Abstract)}},
  author    = {Ciucanu, Radu and Lafourcade, Pascal and Marcadet, Gaël and Soare, Marta},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2023},
  pages     = {6863-6867},
  doi       = {10.24963/IJCAI.2023/772},
  url       = {https://mlanthology.org/ijcai/2023/ciucanu2023ijcai-samba/}
}