Over-the-Air Ensemble Inference with Model Privacy
This work addresses efficient and private inference for edge computing applications, representing an incremental improvement in wireless communication techniques.
The paper tackled the problem of distributed inference at the wireless edge by proposing over-the-air ensemble methods that improve accuracy and model privacy, showing significant performance gains over orthogonal methods with less resource usage and privacy guarantees.
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.