- Generators
graphlet
eigen
more ...
Selectors
SeFilterConstant
gbdt
Graph
netlsd
NxAverageClustering
more ... |
+ Generators
Graphlets
EigenGNN
more ...
Selectors
SeFilterConstant
gbdt
Graph
netlsd
NxAverageClustering
more ... |
Node Classification
GCN
GAT
GraphSAGE
Graph Classification
GIN
TopKPool |
Algorithms
@@ -126,7 +127,7 @@ The documentation will be automatically generated under `docs/_build/html`
## Cite
-You can cite [our paper](https://arxiv.org/abs/2104.04987) as follows if you use this code in your own work:
+Please cite [our paper](https://openreview.net/forum?id=0yHwpLeInDn) as follows if you find our code useful:
```
@inproceedings{
guan2021autogl,
@@ -138,6 +139,16 @@ url={https://openreview.net/forum?id=0yHwpLeInDn}
}
```
-## License
+You may also find our [survey paper](http://arxiv.org/abs/2103.00742) helpful:
+```
+@article{zhang2021automated,
+ title={Automated Machine Learning on Graphs: A Survey},
+ author={Zhang, Ziwei and Wang, Xin and Zhu, Wenwu},
+ booktitle = {Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, {IJCAI-21}},
+ year={2021},
+ note={Survey track}
+}
+```
-We follow [MIT license](LICENSE) across the entire codebase.
+## License
+Notice that we follow [Apache license](LICENSE) across the entire codebase from v0.2.
diff --git a/tune_results.md b/tune_results.md
deleted file mode 100644
index 777d909..0000000
--- a/tune_results.md
+++ /dev/null
@@ -1,62 +0,0 @@
-# Help tune the hp space of models
-
-## New Feature
-
-#### 2020.09.25
-
-Now you can use multiprocessing to quickly get the final result. Seems quicker than `nodeclf_explore_hp_space.py`.
-```
-python nodeclf_hp_multiprocessing.py --dataset xxx xxx --exclude_device 3 4 5 --max_evals 1 5 10 20 50 --path_to_config ../configs/your_config.yaml --hpo random (or tpe) --cluster32
-```
-You can use exclude_device to ignore the devices that you can see in `NVIDIA-SMI`.
-
-**attention!**
-DO NOT USE CUDA_VISIBLE_DEVICES=xx to force the devices, which will be of no effect when running this code.
-
-**attention!**
-If you run your code on cluster32, you must add `--cluster32` since the GPU mapping in cluster32 is a bit strange.
-
-## Environment
-- Python >= 3.6.0
-- PyTorch >= 1.5.1
-- PyTorch-Geometric >= 1.5.1
-- scikit-learn >= 0.23.2
-- tabulate >= 0.8.7
-- lightgbm >= 2.3.0
-- psutil >= 5.7.2
-- hyperopt >= 0.2.4
-- chocolate >= 0.0.2
-- bayesian-optimization >= 1.2.0
-- NetLSD >= 1.0.2
-
-## How to tune
-
-### 1. change the model and hp space
-
-You need to change certain lines in python file `examples/graphclf_explore_hp_space.py` or `examples/nodeclf_explore_hp_space.py` to set your models and adjust your hp spaces. The hp space example is given at the beginning of the files.
-
-### 2. run the code
-
-```
-cd examples
-python xxx_explore_hp_space.py --dataset d1 d2 d3 ... --max_evals max_hpo_eval_time
-```
-Run the cmd above to get randomly sampled `max_hpo_eval_time` models out of your search space on the specifid dataset.
-
-### 3. analyze the results
-The step 2 will give you the models with results. You need to adjust the hp space in step 1 according to the results.
-
-To help simplify the analyze process, you can also leverage the analyze tools under `analysis` folder.
-
-First, copy and paste the results (all the lines with the format `{*some hp} score`. Refer to `analysis/record.txt` for an example) to `analysis/record.txt`, then run:
-```
-python ./analysis/analysis_result.py --path ./analysis/record.txt
-```
-
-This will sort the results for you, so that you can observe the common problems bad models have and regularize the corresponding hp space.
-
-### 4. update log
-
-2020.09.03 Now the re-factored models are merged from branch `lhy_graph_clf`.
-
-2020.09.06 Update all modules from master, add graph explore support.
\ No newline at end of file
|
-AutoGL uses `datasets` to maintain dataset for graph-based machine learning, which is based on Dataset in PyTorch Geometric with some support added to corporate with the auto solver framework.
+AutoGL uses `datasets` to maintain datasets for graph-based machine learning, which is based on Dataset in PyTorch Geometric with some functions added to support the auto solver framework.
-Different graph-based machine learning tasks are solved by different `AutoGL solvers`, which make use of five main modules to automatically solve given tasks, namely `auto feature engineer`, `neural architecture search`, `auto model`, `hyperparameter optimization`, and `auto ensemble`.
+Different graph-based machine learning tasks are handled by different `AutoGL solvers`, which make use of five main modules to automatically solve given tasks, namely `auto feature engineer`, `neural architecture search`, `auto model`, `hyperparameter optimization`, and `auto ensemble`.
Currently, the following algorithms are supported in AutoGL:
@@ -40,7 +41,7 @@ Currently, the following algorithms are supported in AutoGL: