hujilin1229 / GraphCompletion

#Stochastic Weight Completion

This is a TensorFlow implementation of Stochastic Weight Completion for Road Networks, as described in our paper:

Jilin Hu, Chenjuan Guo, Bin Yang, Christian S. Jensen, Stochastic Weight Completion for Road Networks Using Graph Convolutional Networks (ICDE 2019)

Requirements

• tensorflow (>0.12)

Run the demo

python gcrn_main_gcnn.py

Data

In order to use your own data, you have to provide

• an N by N adjacency matrix (N is the number of nodes),
• an N by D feature matrix (D is the number of features per node), and

Have a look at the BatchLoader class in utils.py for an example.

You can specify a dataset as follows:

python gcrn_main_gcnn.py --server_name='chengdu'

(or by editing gcrn_main_gcnn.py)

Cite

Please cite our paper if you use this code in your own work:

@inproceedings{hu2019stochastic,
  title={Stochastic Weight Completion for Road Networks using Graph Convolutional Networks},
  author={Hu, Jilin and Guo, Chenjuan and Yang, Bin and Jensen, Christian S},
  booktitle={2019 IEEE 35th International Conference on Data Engineering (ICDE)},
  pages={1274--1285},
  year={2019},
  organization={IEEE}
}