uTile PET: a Privacy Preserving Solution for Collaborative Data-Driven Projects

EasyChair Preprint 5354

Abstract

There are an increasing number of data-driven space projects. In these projects, the solution is as important as the quality of the data. Moreover, the more data available, the better the performance, so it is normal for different entities to collaborate on a common solution. However, this can be a problem in terms of privacy and it may not always be possible to share data between the different parties. Therefore, we present uTile PET, a solution for the collaborative development of Artificial Intelligence algorithms without the need to compromise the privacy of each of the parties. In addition, SHK-means, a clustering algorithm that works with distributed data and maintains privacy at all times, is presented as an example.

Keyphrases: Federated Learning, K-means clustering, multi-party computation, privacy preserving

@booklet{EasyChair:5354,
  author    = {Daniel Hurtado Ramírez and Luis Porras Díaz and Álvaro Calzado Pérez and Borja Irigoyen Peña and Alexander Benítez Buenache and Juan Miguel Auñón García and Ana María García Sánchez and Pablo González Fuente},
  title     = {uTile PET: a Privacy Preserving Solution for Collaborative Data-Driven Projects},
  howpublished = {EasyChair Preprint 5354},
  year      = {EasyChair, 2021}}