dmlc--dgl
xuyou314
85c9ff01c9
* [Example] Finish adaptive sampling * refactor code and use argparse for command-line usage * Update README.md * add node_per_layer command-line argument
13 行
472 B
Markdown
13 行
472 B
Markdown
# Adaptive sampling for graph representation learning
|
|
|
|
This is dgl implementation of [Adaptive Sampling Towards Fast Graph Representation Learning](https://arxiv.org/abs/1809.05343).
|
|
|
|
The authors' implementation can be found [here](https://github.com/huangwb/AS-GCNN).
|
|
|
|
## Performance
|
|
|
|
Test accuracy on cora dataset achieves 0.84 around 250 epochs when sample size is set to 256 for each layer.
|
|
|
|
## Usage
|
|
|
|
`python adaptive_sampling.py --batch_size 20 --node_per_layer 40` |