View source for Beijing University of Posts and Telecommunications (20240232719). SEMI-FEDERATED LEARNING METHOD BASED ON NEXT-GENERATION MULTIPLE ACCESS TECHNOLOGY simplified abstract

=SEMI-FEDERATED LEARNING METHOD BASED ON NEXT-GENERATION MULTIPLE ACCESS TECHNOLOGY=


==Organization Name==

[[:Category:Beijing University of Posts and Telecommunications|Beijing University of Posts and Telecommunications]]


[[Category:Beijing University of Posts and Telecommunications]]

==Inventor(s)==

[[:Category:Hui Tian of Beijing (CN)|Hui Tian of Beijing (CN)]][[Category:Hui Tian of Beijing (CN)]]

[[:Category:Wanli Ni of Beijing (CN)|Wanli Ni of Beijing (CN)]][[Category:Wanli Ni of Beijing (CN)]]

[[:Category:Ping Zhang of Beijing (CN)|Ping Zhang of Beijing (CN)]][[Category:Ping Zhang of Beijing (CN)]]

[[:Category:Keyan Liu of Beijing (CN)|Keyan Liu of Beijing (CN)]][[Category:Keyan Liu of Beijing (CN)]]

[[:Category:Shaoshuai Fan of Beijing (CN)|Shaoshuai Fan of Beijing (CN)]][[Category:Shaoshuai Fan of Beijing (CN)]]

[[:Category:Gaofeng Nie of Beijing (CN)|Gaofeng Nie of Beijing (CN)]][[Category:Gaofeng Nie of Beijing (CN)]]

==SEMI-FEDERATED LEARNING METHOD BASED ON NEXT-GENERATION MULTIPLE ACCESS TECHNOLOGY - A simplified explanation of the abstract==


This abstract first appeared for US patent application 20240232719 titled 'SEMI-FEDERATED LEARNING METHOD BASED ON NEXT-GENERATION MULTIPLE ACCESS TECHNOLOGY

'''Simplified Explanation''': 

This patent application introduces a semi-federated learning method that combines centralized learning with federated learning to allow devices with limited computing capabilities to contribute to training a global model. It utilizes a reconfigurable intelligent surface to dynamically modify the channel environment, enabling different users to meet their specific task requirements simultaneously. By integrating communication-centric and computing-centric users, data transmission efficiency is improved, and the overall accuracy of the global model is enhanced.

'''Key Features and Innovation''':
* Integration of centralized learning and federated learning for improved model training.
* Deployment of a reconfigurable intelligent surface for dynamic channel environment adaptation.
* Simultaneous data transmission by different user types on the same resource.
* Optimization of user power allocation and intelligent surface configuration to reduce power consumption.
* Prolongation of the life cycle of an intelligent Internet of Things network.

'''Potential Applications''':
The technology can be applied in various fields such as telecommunications, IoT networks, machine learning, and data analytics.

'''Problems Solved''':
The technology addresses the limitations of devices with weak computing capabilities in contributing to global model training, as well as the inefficiencies in data transmission and power consumption in heterogeneous user environments.

'''Benefits''':
* Enhanced model accuracy through collaborative learning.
* Improved data transmission efficiency.
* Reduced power consumption for prolonged network operation.

'''Commercial Applications''':
Potential commercial applications include telecommunications infrastructure optimization, IoT network management solutions, and machine learning platform development for various industries.

'''Prior Art''':
Readers interested in prior art related to this technology can explore research on federated learning, intelligent surfaces, and IoT network optimization.

'''Frequently Updated Research''':
Stay updated on the latest advancements in federated learning, intelligent surface technologies, and IoT network optimization for further insights into this field.

'''Questions about Semi-Federated Learning''':
1. How does the integration of centralized learning and federated learning benefit the training of a global model?
2. What are the key advantages of using a reconfigurable intelligent surface in a semi-federated learning system?


==Original Abstract Submitted==

a semi-federated learning (semifl) method based on a next-generation multiple access (ngma) technology is provided. centralized learning (cl) and fl are integrated such that devices with weak computing capabilities can also participate in training of a global model. a simultaneously transmitting and reflecting reconfigurable intelligent surface (star-ris) is deployed to dynamically change a channel environment such that a system can meet different task requirements of heterogeneous users. communication-centric cl users and computing-centric fl users can transmit data in parallel on a same time-frequency resource. this avoids a waste of data resources, enriches data obtaining of a base station (bs), and improves accuracy of the global model. the semifl method also integrates a strategy for jointly optimizing user power allocation and a configuration of the star-ris to reduce total uplink transmit power consumption of the system and prolong a life cycle of an intelligent internet of things (iot) network.



[[Category:G06N20/00]]
[[Category:CPC_G06N20/00]]