Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
Theheavens d71b84d89b | 2 years ago | |
---|---|---|
.. | ||
README.md | 2 years ago |
Paper:Heterogeneous Graph Representation Learning with Relation Awareness
Code from author:https://github.com/yule-BUAA/R-HGNN
Clone the Openhgnn-DGL
python main.py -m RHGNN -t node_classification -d imdb4GTN -g 0 --use_best_config
Candidate dataset: imdb4GTN/acm4GTN
If you do not have gpu, set -gpu -1.
imdb4GTN
NOTE: imdb4GTN is a small-scale dataset. We will add large-scale datasets (OGB-MAG, OAG-Venue and OAG-L1-Field) in our further work.
Node classification
accuracy | |
---|---|
imdb4GTN | 0.5883 |
first step: hetero_conv and residual connection
second step: relation_crossing_layer
third step: relation_propagation_layer
R-HGNN layer is composed of these three components and stack L layers to receive information from multi-hop neighbors. Finally, the L layers could provide relation-aware node representations for target node.
num_heads = 8
hidden_units = 64
relation_hidden_units = 8
n_layers = 2
learning_rate = 0.001
dropout = 0.5
residual = True
Best config can be found in best_config
Tianyu Zhao, Qi Zhang[GAMMA LAB]
Submit an issue or email to 1049371730@qq.com.
OpenHGNN是由北邮GAMMA Lab开发的基于PyTorch和DGL的开源异质图神经网络工具包。
Python Markdown Shell
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》