.png){kind=icon}
.png){kind=icon}
Return
Data-Representation Aware Resource Scheduling for Edge Intelligence
.png){kind=icon}
PRE
AI
DOI：10.1109/TVT.2022.3195571.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
As a large amount of real-time data generated by billions of mobile devices nowadays, how to efficiently upload and process these data to server to realize ubiquitous intelligence at network edge is critical. A key challenge is to use limited radio resources to transmit massive data samples. Motivated by active learning, a communication-efficient way is to select most informative data samples for model training. In this article, we develop a new data scheduling scheme based on data representation. Specifically, the devices would like to transmit the most representative data samples, while the server prefers to schedule the least representative ones. A scheduling scheme is designed for such value asymmetry of data samples. The experimental results demonstrate that the proposed scheme can significantly improve the performance of machine learning.
Keywords:
Data representation
edge intelligence
machine learning
resource allocation
scheduling
Journal
IF:
7.1
Papers:
1.8W
Citations:
6.6W