Communications of the ACM
Technique Smooths Path for 'Federated Learning' AI Training in Wireless Devices
In federated learning, a form of machine learning involving multiple devices, each client is trained on different data and develops its own model for performing a specific task.
Credit: Jonas Leupe
Researchers at North Carolina State University (NC State) have developed a new technique that allows federated learning to be used to train artificial intelligence (AI) systems on wireless devices.
The technique uses compression to reduce the size of data transmissions, which is important given that federated learning requires a substantial amount of communication between the clients and central server during training.
The data packets are condensed before transmission and reconstructed by the centralized server.
The algorithms developed by the researchers were able to condense the amount of wireless data shipped from the clients by upwards of 99%.
Said NC State's Kai Yue, "Our technique makes federated learning viable for wireless devices where there is limited available bandwidth. For example, it could be used to improve the performance of many AI programs that interface with users, such as voice-activated virtual assistants."
From NC State University News
View Full Article
Abstracts Copyright © 2022 SmithBucklin, Washington, DC, USA
No entries found