東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

FindBook

Google Book

Amazon

博客來

Learning from Relational Data via Graph Neural Networks.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Learning from Relational Data via Graph Neural Networks./
作者:	Liu, Linfeng.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2022,
面頁冊數:	167 p.
附註:	Source: Dissertations Abstracts International, Volume: 83-12, Section: B.
Contained By:	Dissertations Abstracts International83-12B.
標題:	Computer science. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29165698
ISBN:	9798438784296

Learning from Relational Data via Graph Neural Networks.
Liu, Linfeng.

Learning from Relational Data via Graph Neural Networks. - Ann Arbor : ProQuest Dissertations & Theses, 2022 - 167 p.

Source: Dissertations Abstracts International, Volume: 83-12, Section: B.

Thesis (Ph.D.)--Tufts University, 2022.

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

Deep learning improves human productivity by automatically learning from massive data and efficiently providing predictions for decision making. Yet it is still challenging to apply deep learning to data with irregular structures. A prominent type of irregular data is relational data, which describes relationships among objects and covers a wide range of applications. However, applying deep learning to relational data is difficult mainly due to the combinatorial nature of relational data. This thesis contributes new methods for the analysis of relational data from several important aspects. We use Graph Neural Networks (GNNs) as primary tools to automatically learn representations that capture important information from relational data. We identify three research domains that either implicitly or explicitly present relational data as graphs, and we reveal the great benefits of adopting curated GNNs to these domains. First, we design graph learning algorithm for Gaussian Process (GP) inference. We impose graphs over data points with an emphasis that the inference of a data point focuses on values at its nearby data points, and then treat GP inference as a graph learning problem. Second, we model label correlations among nearby data points where the data points are not i.i.d. distributed. Our model combines the advantages of both representation learning and probabilistic methods to better encode correlations in relational data. We show promising results in spatial data modeling and graph node classification. Third, we learn graph representations to solve graph-related combinatorial problems. The model developed is capable of utilizing the knowledge from solved problem instances to accelerate the procedure of solving future instances. We explore the usage of probabilistic methods to enable efficient optimization. Our model improves performance over previous models. These three domains studied in this thesis demonstrate the power of combining probabilistic methods and deep learning in modeling relational data for solving real-world problems.

ISBN: 9798438784296Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

Relational data

Learning from Relational Data via Graph Neural Networks.
LDR:03091nmm a2200325 4500 001 2350411
005 20221020125757.5
008 241004s2022 ||||||||||||||||| ||eng d
020 $a 9798438784296
035 $a (MiAaPQ)AAI29165698
035 $a AAI29165698
040 $a MiAaPQ $c MiAaPQ
100 1 $a Liu, Linfeng. $3 3488635
245 1 0 $a Learning from Relational Data via Graph Neural Networks.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2022
300 $a 167 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-12, Section: B.
500 $a Advisor: Liu, Liping.
502 $a Thesis (Ph.D.)--Tufts University, 2022.
506 $a This item must not be sold to any third party vendors.
520 $a Deep learning improves human productivity by automatically learning from massive data and efficiently providing predictions for decision making. Yet it is still challenging to apply deep learning to data with irregular structures. A prominent type of irregular data is relational data, which describes relationships among objects and covers a wide range of applications. However, applying deep learning to relational data is difficult mainly due to the combinatorial nature of relational data. This thesis contributes new methods for the analysis of relational data from several important aspects. We use Graph Neural Networks (GNNs) as primary tools to automatically learn representations that capture important information from relational data. We identify three research domains that either implicitly or explicitly present relational data as graphs, and we reveal the great benefits of adopting curated GNNs to these domains. First, we design graph learning algorithm for Gaussian Process (GP) inference. We impose graphs over data points with an emphasis that the inference of a data point focuses on values at its nearby data points, and then treat GP inference as a graph learning problem. Second, we model label correlations among nearby data points where the data points are not i.i.d. distributed. Our model combines the advantages of both representation learning and probabilistic methods to better encode correlations in relational data. We show promising results in spatial data modeling and graph node classification. Third, we learn graph representations to solve graph-related combinatorial problems. The model developed is capable of utilizing the knowledge from solved problem instances to accelerate the procedure of solving future instances. We explore the usage of probabilistic methods to enable efficient optimization. Our model improves performance over previous models. These three domains studied in this thesis demonstrate the power of combining probabilistic methods and deep learning in modeling relational data for solving real-world problems.
590 $a School code: 0234.
650 4 $a Computer science. $3 523869
650 4 $a Artificial intelligence. $3 516317
653 $a Relational data
653 $a Graph neural networks
690 $a 0984
690 $a 0800
710 2 $a Tufts University. $b Computer Science. $3 1026183
773 0 $t Dissertations Abstracts International $g 83-12B.
790 $a 0234
791 $a Ph.D.
792 $a 2022
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29165698