請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/83187
標題: | 利用深度學習方法繪製動態圖形 A Deep Learning Approach for Dynamic Graph Drawing |
其他標題: | A Deep Learning Approach for Dynamic Graph Drawing |
作者: | 王思涵 Sih-Han Wang |
指導教授: | 顏嗣鈞 Hsu-Chun Yen |
關鍵字: | 動態圖形繪製,圖佈局,使用者偏好,深度學習,圖神經網路, Dynamic graph drawing,Graph layout,Human preference,Deep learning,Graph Neural Network, |
出版年 : | 2022 |
學位: | 碩士 |
摘要: | 圖形視覺化可以幫助用戶理解或分析網絡資訊。已經有許多傳統算法被提出來可視化圖形。如今,有很多深度學習技術也被應用於圖形繪製。然而,目前還沒有關於使用深度神經網絡繪製動態圖的相關工作。出於這個原因,我們想擴展現有的工作來探索動態圖形繪製的可能性。因此,我們提出了一種用於動態圖形繪製的深度學習方法。 該方法可以使用圖形神經網絡同時優化動態圖形繪製的多種美學。我們可以選擇任何我們想要優化的指標,以獲得更好的動態圖形可視化結果。此外,我們可以決定我們對特定美學的偏好以獲得所需的佈局。在實驗中,我們將詳細報告動態圖可視化的結果,我們還將討論本文的一些局限性和未來的工作。 Graph visualization provides a useful tool for helping users to better reveal or analyze the network information associated with an application. Many traditional algorithms have been proposed to visualize graphs. Although there are some deep learning techniques for graph drawing, as far as we know there is no prior work on dynamic graph drawing using deep neural networks. For this reason, we want to extend the existing work to explore the possibility of dynamic graph drawing. Therefore, we propose a deep learning approach for dynamic graph drawing. The approach uses graph neural networks to optimize multiple aesthetics for dynamic graph drawing simultaneously. We could choose any metrics we want to optimize for better dynamic graph visualization. In addition, we could decide our preference for specific aesthetics to get the desired layouts. In the experiment, we report the results of dynamic graph visualization in detail, and we also discuss some limitations and future work. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/83187 |
DOI: | 10.6342/NTU202203955 |
全文授權: | 同意授權(限校園內公開) |
顯示於系所單位: | 電機工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
U0001-2409202202425300.pdf 授權僅限NTU校內IP使用(校園外請利用VPN校外連線服務) | 1.5 MB | Adobe PDF | 檢視/開啟 |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。