Published February 7, 2022 | Version v2
Conference paper Open

Over-the-Air Ensemble Inference with Model Privacy

  • 1. Imperial College London

Description

We consider distributed inference at the wireless edge, where multiple clients with an ensemble of models, each trained independently on a local dataset, are queried in parallel to make an accurate decision on a new sample. In addition to maximizing inference accuracy, we also want to maximize the privacy of local models. We exploit the superposition property of the air to implement bandwidth-efficient ensemble inference methods. We introduce different over-the-air ensemble methods and show that these schemes perform significantly better than their orthogonal counterparts, while using less resources and providing privacy guarantees. We also provide experimental results verifying the benefits of the proposed over-the-air inference approach, whose source code is shared publicly on Github.

Notes

To appear in IEEE International Symposium on Information Theory (ISIT) 2022

Files

Over-the-Air Ensemble Inference with Model Privacy.pdf

Files (872.7 kB)

Additional details

Funding

GREENEDGE – Taming the environmental impact of mobile networks through GREEN EDGE computing platforms 953775
European Commission