Efficient Privacy Preserving Neural Network Inference for Heart Arrythmia Detection
Loading...
Author
Chervet, Philipp
Olteanu, Alexandra-Mihaela
Troncoso-Pastoriza, Juan Ramón
Froelicher, David
Van Zaen, Jérôme
Delgado-Gonzalo, Ricard
Hubaux, Jean-Pierre
DOI
Abstract
The raise of AI and machine learning as a service (MLaaS) poses a risk to the privacy of those using it. In today’s data-driven application landscape, it is common that a party needs to process sensitive (personal) data using third party resources (such as computation, storage, or communication infrastructure), which constitutes a risk with respect to the privacy of such data. CSEM is working on performing neural network (NN) inference without revealing user input data to other parties involved and while hiding the model parameters from the user. Relying on homomorphic encryption and secure two-party computation, we present here a service as a client-server application for privacy preserving NN inference.
Publication Reference
CSEM Scientific and Technical Report 2019, p. 96
Year
2019