Hongyi ZHANG 89fd3285b7 | 1 year ago | |
---|---|---|
.. | ||
emb | 2 years ago | |
CompGCN.py | 2 years ago | |
DMGI.py | 2 years ago | |
GATNE.py | 2 years ago | |
GTN_sparse.py | 2 years ago | |
HAN.py | 2 years ago | |
HDE.py | 2 years ago | |
HGAT.py | 1 year ago | |
HGConv.py | 2 years ago | |
HGNN_AC.py | 2 years ago | |
HGSL.py | 2 years ago | |
HGT.py | 2 years ago | |
HGT_hetero.py | 2 years ago | |
HPN.py | 2 years ago | |
HeCo.py | 2 years ago | |
HeGAN.py | 2 years ago | |
HetGNN.py | 2 years ago | |
HetSANN.py | 1 year ago | |
HeteroMLP.py | 2 years ago | |
KGCN.py | 2 years ago | |
MAGNN.py | 2 years ago | |
MHNF.py | 2 years ago | |
Micro_layer.py | 2 years ago | |
Multi_level.py | 2 years ago | |
NARS.py | 2 years ago | |
NEW_model.py | 3 years ago | |
NSHE.py | 2 years ago | |
README.md | 2 years ago | |
RGAT.py | 2 years ago | |
RGCN.py | 2 years ago | |
RHGNN.py | 2 years ago | |
RSHN.py | 2 years ago | |
SLiCE.py | 1 year ago | |
SimpleHGN.py | 1 year ago | |
SkipGram.py | 2 years ago | |
TransD.py | 2 years ago | |
TransE.py | 2 years ago | |
TransH.py | 2 years ago | |
TransR.py | 2 years ago | |
__init__.py | 1 year ago | |
base_model.py | 2 years ago | |
fastGTN.py | 2 years ago | |
general_HGNN.py | 2 years ago | |
homo_GNN.py | 2 years ago | |
ieHGCN.py | 1 year ago |
A model plays a role of encoder. For input given a hetero-graph or MFG and feature (if available), the model need to output a dict of node embedding. In general, it should ouput all nodes embedding. It is allowed that just output the embedding of target nodes which are participated in loss calculation.
It mainly contains two parts: model builder and forward propagation.
We create a classmethod build_model_from_args for every model. with that, we can use args and hg to build up a custom model with model-specific hyper-parameters. So it is necessary to implement the func build_model_from_args in your model.
def forward(self, hg=None, feat_dict=None, *args, **wkags):
'''
Parameters
----------
g : dgl.heterograph
the dgl heterogeneous graph
feat_dict : dict
the feature matrix dict of different node types,
e.g {'author':feat_of_author[number of author nodes * dimension], ...}
Returns
-------
out_dict: dict
the same with feat_dict
'''
return out_dict
Input: The args hg and feat_dict are not necessary.
Output: A dict of embedding.
Model | Embedding | GNN | Knowledge Graph |
---|---|---|---|
Metapath2vec[KDD 2017] | ✔️ | ||
RGCN[ESWC 2018] | ✔️ | ✔️ | |
HAN[WWW 2019] | ✔️ | ||
HetGNN[KDD 2019] | ✔️ | ✔️ | |
GTN[NeurIPS 2019] | ✔️ | ||
RSHN[ICDM 2019] | ✔️ | ||
MAGNN[WWW 2020] | ✔️ | ✔️ | |
NSHE[IJCAI 2020] | ✔️ | ✔️ | |
CompGCN[ICLR 2020] | ✔️ |
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》