Federated Pruning: Improving Neural Network Efficiency with Federated Learning

TL;DR

This paper introduces Federated Pruning, a method to reduce neural network size in federated learning for speech recognition, maintaining performance while addressing resource constraints and leveraging client data.

Contribution

It proposes a novel federated pruning approach that improves model efficiency and performance by utilizing decentralized data and exploring various pruning schemes.

Findings

Federated pruning achieves comparable accuracy to full models.

Pruning benefits are enhanced by leveraging client data.

The method reduces model size and communication costs.

Abstract

Automatic Speech Recognition models require large amount of speech data for training, and the collection of such data often leads to privacy concerns. Federated learning has been widely used and is considered to be an effective decentralized technique by collaboratively learning a shared prediction model while keeping the data local on different clients devices. However, the limited computation and communication resources on clients devices present practical difficulties for large models. To overcome such challenges, we propose Federated Pruning to train a reduced model under the federated setting, while maintaining similar performance compared to the full model. Moreover, the vast amount of clients data can also be leveraged to improve the pruning results compared to centralized training. We explore different pruning schemes and provide empirical evidence of the effectiveness of our methods.