HPN[TKDE2021]
Paper: Heterogeneous Graph Propagation Network
The author did not provide codes. So, we complete it according to the implementation of HAN
How to run
Clone the Openhgnn-DGL
python main.py -m HPN -t node_classification -d acm_han_raw -g 0
If you do not have gpu, set -gpu -1.
Candidate dataset: acm_han_raw, acm4GTN, imdb4GTN
Performance
HPN[OpenHGNN] |
Macro-F1 |
Micro-F1 |
acm_han_raw |
91.80 |
91.80 |
acm4GTN |
91.04 |
90.92 |
imdb4GTN |
60.96 |
64.00 |
TrainerFlow: node_classification
The model is trained in semi-supervisied node classification.
model
HPNLayer
SemanticFusion
Dataset
Supported dataset: acm_han_raw, acm4GTN, imdb4GTN
You can download the dataset by
wget https://s3.cn-north-1.amazonaws.com.cn/dgl-data/dataset/acm_han_raw.zip
wget https://s3.cn-north-1.amazonaws.com.cn/dgl-data/dataset/acm4GTN.zip
wget https://s3.cn-north-1.amazonaws.com.cn/dgl-data/dataset/imdb4GTN.zip
Hyper-parameter specific to the model
k_layer = 1 # iterations in Semantic Propagation
alpha = 0.1 # Value of restart probability
out_embedsize = 64 # Dimension of the final embedding
edge_drop = 0 # the dropout rate on edges that controls the messages received by each node
Best config can be found in best_config
Related API in DGL
dgl.metapath_reachable_graph
APPNPConv
More
Contirbutor
Donglin Xia[GAMMA LAB]
If you have any questions,
Submit an issue or email to donglin.xia@bupt.edu.cn.