Privatized graph federated learning
Abstract Federated learning is a semi-distributed algorithm, where a server communicates with multiple dispersed clients to learn a global model. The federated architecture is not robust and is sensitive to communication and computational overloads due to its one-master multi-client structure. It ca...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2023-08-01
|
| Series: | EURASIP Journal on Advances in Signal Processing |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s13634-023-01049-4 |
| _version_ | 1797451031505797120 |
|---|---|
| author | Elsa Rizk Stefan Vlaski Ali H. Sayed |
| author_facet | Elsa Rizk Stefan Vlaski Ali H. Sayed |
| author_sort | Elsa Rizk |
| collection | DOAJ |
| description | Abstract Federated learning is a semi-distributed algorithm, where a server communicates with multiple dispersed clients to learn a global model. The federated architecture is not robust and is sensitive to communication and computational overloads due to its one-master multi-client structure. It can also be subject to privacy attacks targeting personal information on the communication links. In this work, we introduce graph federated learning, which consists of multiple federated units connected by a graph. We then show how graph-homomorphic perturbations can be used to ensure the algorithm is differentially private on the server level. While on the client level, we show that improvement in the differentially private federated learning algorithm can be attained through the addition of random noise to the updates, as opposed to the models. We conduct both convergence and privacy theoretical analyses and illustrate performance by means of computer simulations. |
| first_indexed | 2024-03-09T14:49:05Z |
| format | Article |
| id | doaj.art-28e354fb47f04b87b5d8702b7703b8b2 |
| institution | Directory Open Access Journal |
| issn | 1687-6180 |
| language | English |
| last_indexed | 2024-03-09T14:49:05Z |
| publishDate | 2023-08-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | EURASIP Journal on Advances in Signal Processing |
| spelling | doaj.art-28e354fb47f04b87b5d8702b7703b8b22023-11-26T14:33:10ZengSpringerOpenEURASIP Journal on Advances in Signal Processing1687-61802023-08-012023113110.1186/s13634-023-01049-4Privatized graph federated learningElsa Rizk0Stefan Vlaski1Ali H. Sayed2School of Engineering, École Polytechnique Fédérale de LausanneDepartment of Electrical and Electronic Engineering, Imperial College LondonSchool of Engineering, École Polytechnique Fédérale de LausanneAbstract Federated learning is a semi-distributed algorithm, where a server communicates with multiple dispersed clients to learn a global model. The federated architecture is not robust and is sensitive to communication and computational overloads due to its one-master multi-client structure. It can also be subject to privacy attacks targeting personal information on the communication links. In this work, we introduce graph federated learning, which consists of multiple federated units connected by a graph. We then show how graph-homomorphic perturbations can be used to ensure the algorithm is differentially private on the server level. While on the client level, we show that improvement in the differentially private federated learning algorithm can be attained through the addition of random noise to the updates, as opposed to the models. We conduct both convergence and privacy theoretical analyses and illustrate performance by means of computer simulations.https://doi.org/10.1186/s13634-023-01049-4Federated learningDistributed learningPrivatized learningDifferntial privacy |
| spellingShingle | Elsa Rizk Stefan Vlaski Ali H. Sayed Privatized graph federated learning EURASIP Journal on Advances in Signal Processing Federated learning Distributed learning Privatized learning Differntial privacy |
| title | Privatized graph federated learning |
| title_full | Privatized graph federated learning |
| title_fullStr | Privatized graph federated learning |
| title_full_unstemmed | Privatized graph federated learning |
| title_short | Privatized graph federated learning |
| title_sort | privatized graph federated learning |
| topic | Federated learning Distributed learning Privatized learning Differntial privacy |
| url | https://doi.org/10.1186/s13634-023-01049-4 |
| work_keys_str_mv | AT elsarizk privatizedgraphfederatedlearning AT stefanvlaski privatizedgraphfederatedlearning AT alihsayed privatizedgraphfederatedlearning |