.png){kind=icon}
.png){kind=icon}
Return
CPGCN: Evolutionary Relationship Mining Framework for Ethnic Costume Elements
.png){kind=icon}
PRE
AI
DOI：10.1007/978-981-95-5719-6_36.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Ethnic costumes serve as a significant medium for material and cultural exchange and integration among different ethnic groups. The correlations between their elements and the evolutionary patterns are of great importance in uncovering the historical connotations of ethnic fusion. However, existing research has predominantly focused on classification and recognition tasks, with limited exploration into the deep interconnections among elements and their implications for ethnic integration. To solve this problem, we propose a new framework for mining the evolutionary relationships of ethnic costume elements, namely Community-Preserving Graph Convolutional Network (CPGCN). This end-to-end framework adaptively integrates link information and attribute information while specifically considering community structures. By predicting relationships between nodes in the feature network of ethnic costume elements, it obtains characteristic foundations of ethnic integration processes. Furthermore, we propose a new ethnic costume element extraction model, Xception-DL. By mining the features of various parts in the ethnic costume element data and calculating the similarity between each feature node, a feature network of ethnic costume elements is constructed. Extensive experiments on multiple common datasets show that CPGCN significantly outperforms existing methods.
Keywords:
Link Prediction
Feature Extraction
Graph Neural Network
Journal
W
IF:
0
Papers:
32
Citations:
0