The first step in setting a Federated Learning project using vantage6 is understanding how it works. Having a general idea of how its components are built and how they are interconnected can make the technical setup much easier for all the involved parties.
To this end, we wrote a paper where we describe the niche that vantage6 fills in the Federated Learning landscape, its conceptual design, and its architecture. You can find the paper here:
This work was presented as a contribution during the AMIA 2020 Virtual Annual Symposium. It was accompanied by an oral presentation, which you can watch right here as well (~9 min, in English):
The Petronas release of vantage6 (v 3.0) had several improvements, the most important being allowing algorithms running at each party to communicate with one another through a virtual private network. This allows the execution of different types of algorithms (e.g., multi-party computation) that were practically infeasible before. This release was presented at the International Conference of Informatics, Management, and Technology in Healthcare (ICIMTH) and comes with its accompanying paper, which you can get right here:
If you are using vantage6, please cite the papers as well as their corresponding GitHub repository as follows:
- Arturo Moncada-Torres, Frank Martin, Melle Sieswerda, Johan van Soest, Gijs Geleijnse. "VANTAGE6: an open source priVAcy preserviNg federaTed leArninG infrastructurE for Secure Insight eXchange". AMIA Annual Symposium Proceedings, 2020, p. 870-877. [BibTeX]
- Djura Smits*, Bart van Beusekom*, Frank Martin, Lourens Veen, Gijs Geleijnse, Arturo. Moncada-Torres, “An Improved Infrastructure for Privacy-Preserving Analysis of Patient Data”, Proceedings of the International Conference of Informatics, Management, and Technology in Healthcare (ICIMTH), vol. 25, 2022, p. 144-147 . [BibTeX]
- Frank Martin, Melle Sieswerda, Hasan Alradhi, et al., vantage6. Available at https://doi.org/10.5281/zenodo.3686944. Accessed on [MONTH, 20XX].
[Asterisks (*) denote either joint first- or last-authorship]