國立虎尾科技大學 |

Joint DNN Partitioning and Resource Allocation for Multiple Machine Learning-based Mobile Applications at the Network Edge.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Joint DNN Partitioning and Resource Allocation for Multiple Machine Learning-based Mobile Applications at the Network Edge./
作者:	Cheng, Cheng-Yu.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2024,
面頁冊數:	103 p.
附註:	Source: Dissertations Abstracts International, Volume: 85-06, Section: B.
Contained By:	Dissertations Abstracts International85-06B.
標題:	Computer engineering. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30815625
ISBN:	9798381163711

Joint DNN Partitioning and Resource Allocation for Multiple Machine Learning-based Mobile Applications at the Network Edge.
Cheng, Cheng-Yu.

Joint DNN Partitioning and Resource Allocation for Multiple Machine Learning-based Mobile Applications at the Network Edge. - Ann Arbor : ProQuest Dissertations & Theses, 2024 - 103 p.

Source: Dissertations Abstracts International, Volume: 85-06, Section: B.

Thesis (Ph.D.)--The Catholic University of America, 2024.

This item must not be sold to any third party vendors.

 The increasing use of artificial intelligence (AI) and machine learning (ML) in mobile applications requires stringent quality of service such as low latency and high accuracy. Deep Neural Networks (DNNs) are the core technique used in AI and ML, but they are computation-intensive and power-consuming, leading to limitations in computing capability and energy consumption on user mobile devices (MD). As a result, many AI/ML applications offload DNN processing tasks from MDs to mobile edge computing servers (ES) or cloud data centers. However, this approach puts significant computational pressure on the servers and requires significant amounts of data to be transmitted over wireless networks. DNNs consist of multiple connected layers of neurons, and the computation complexity and amount of intermediate output data vary at each layer. Partitioning the DNN inference operations between mobile devices and edge/cloud servers based on the DNN layer characteristics can significantly reduce computation pressure, power, and data rate requirements at the mobile device, while improving overall operation efficiency, latency, accuracy, and privacy, that is, the optimal DNN partitioning point and offloading strategy should depend on the computation and output data size characteristics of the layer compositions, available resources at the mobile devices and servers, and network environments, especially, when there are multiple concurrent DNN operations by multiple mobile users along with multiple edge servers in a practical mobile edge computing (MEC) network. Thus, there is a need for a framework to optimize performance of multiple DNN operations with multiple mobile devices and multiple edge servers in a MEC network.

ISBN: 9798381163711Subjects--Topical Terms:

569006
Computer engineering.
Subjects--Index Terms:

Machine learning

Joint DNN Partitioning and Resource Allocation for Multiple Machine Learning-based Mobile Applications at the Network Edge.
LDR:03026nam a2200397 4500 001 1121928
005 20240624103715.5
006 m o d
007 cr#unu||||||||
008 240823s2024 ||||||||||||||||| ||eng d
020 $a 9798381163711
035 $a (MiAaPQ)AAI30815625
035 $a AAI30815625
040 $a MiAaPQ $c MiAaPQ
100 1 $a Cheng, Cheng-Yu. $0 (orcid)0000-0002-1939-3268 $3 1437830
245 1 0 $a Joint DNN Partitioning and Resource Allocation for Multiple Machine Learning-based Mobile Applications at the Network Edge.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2024
300 $a 103 p.
500 $a Source: Dissertations Abstracts International, Volume: 85-06, Section: B.
500 $a Advisor: Liu, Hang.
502 $a Thesis (Ph.D.)--The Catholic University of America, 2024.
506 $a This item must not be sold to any third party vendors.
520 $a  The increasing use of artificial intelligence (AI) and machine learning (ML) in mobile applications requires stringent quality of service such as low latency and high accuracy. Deep Neural Networks (DNNs) are the core technique used in AI and ML, but they are computation-intensive and power-consuming, leading to limitations in computing capability and energy consumption on user mobile devices (MD). As a result, many AI/ML applications offload DNN processing tasks from MDs to mobile edge computing servers (ES) or cloud data centers. However, this approach puts significant computational pressure on the servers and requires significant amounts of data to be transmitted over wireless networks. DNNs consist of multiple connected layers of neurons, and the computation complexity and amount of intermediate output data vary at each layer. Partitioning the DNN inference operations between mobile devices and edge/cloud servers based on the DNN layer characteristics can significantly reduce computation pressure, power, and data rate requirements at the mobile device, while improving overall operation efficiency, latency, accuracy, and privacy, that is, the optimal DNN partitioning point and offloading strategy should depend on the computation and output data size characteristics of the layer compositions, available resources at the mobile devices and servers, and network environments, especially, when there are multiple concurrent DNN operations by multiple mobile users along with multiple edge servers in a practical mobile edge computing (MEC) network. Thus, there is a need for a framework to optimize performance of multiple DNN operations with multiple mobile devices and multiple edge servers in a MEC network.
590 $a School code: 0043.
650 4 $a Computer engineering. $3 569006
650 4 $a Computer science. $3 573171
653 $a Machine learning
653 $a Deep Neural Networks
653 $a Mobile devices
653 $a Mobile edge computing
653 $a Operation efficiency
690 $a 0984
690 $a 0464
690 $a 0800
710 2 $a The Catholic University of America. $b Electrical Engineering and Computer Science. $3 1180189
773 0 $t Dissertations Abstracts International $g 85-06B.
790 $a 0043
791 $a Ph.D.
792 $a 2024
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=30815625