17941121. HARDWARE-AWARE FEDERATED LEARNING simplified abstract (QUALCOMM Incorporated)

HARDWARE-AWARE FEDERATED LEARNING

Organization Name

QUALCOMM Incorporated

Inventor(s)

An Chen of San Diego CA (US)

Vijaya Datta Mayyuri of San Diego CA (US)

HARDWARE-AWARE FEDERATED LEARNING - A simplified explanation of the abstract

This abstract first appeared for US patent application 17941121 titled 'HARDWARE-AWARE FEDERATED LEARNING

Simplified Explanation

The abstract describes a method for hardware-aware federated learning, where a device receives information on a jointly-trained artificial neural network (ANN) from a server, determines its current hardware capability for on-device training, and transmits this information back to the server. The server then sends an adapted version of the ANN based on the device's hardware capability.

• Device receives information on a jointly-trained ANN from server
• Determines current hardware capability for on-device training
• Transmits hardware capability information to server
• Receives adapted version of ANN based on hardware capability from server

Potential Applications

• Personalized on-device training of artificial neural networks
• Adaptive learning models based on device hardware capabilities

Problems Solved

• Optimizing training of neural networks for different devices
• Improving efficiency and performance of on-device training

Benefits

• Customized training models for individual devices
• Enhanced performance and efficiency in on-device training
• Adaptation to varying hardware capabilities for improved learning outcomes

Original Abstract Submitted

A processor-implemented method for hardware-aware federated learning includes receiving, from a server, information corresponding to a first jointly-trained artificial neural network (ANN). A current hardware capability of a device for on-device training of the first jointly-trained ANN is determined. The device transmits an indication of the current hardware capability to the server. In response to the transmitted indication, the device receives information corresponding to a second jointly-trained ANN) from the server. The second jointly-trained ANN is an adapted version of the first jointly-trained ANN generated based on the indication of the current hardware capability.