# Wireless Federated Distillation for Distributed Edge Learning with Heterogeneous Data

## TL;DR

This paper explores wireless implementations of federated learning and distillation, proposing digital and over-the-air schemes to address communication challenges in distributed edge learning with heterogeneous data.

## Contribution

It introduces a novel Hybrid Federated Distillation scheme and evaluates digital and over-the-air implementations over Gaussian channels.

## Key findings

- Digital and over-the-air schemes are effective over Gaussian channels.
- Hybrid Federated Distillation improves communication efficiency.
- Wireless implementations face unique challenges and opportunities.

## Abstract

Cooperative training methods for distributed machine learning typically assume noiseless and ideal communication channels. This work studies some of the opportunities and challenges arising from the presence of wireless communication links. We specifically consider wireless implementations of Federated Learning (FL) and Federated Distillation (FD), as well as of a novel Hybrid Federated Distillation (HFD) scheme. Both digital implementations based on separate source-channel coding and over-the-air computing implementations based on joint source-channel coding are proposed and evaluated over Gaussian multiple-access channels.