Merge branch 'PaddlePaddle:main' into pna

examples/pna/README.md

@@ -0,0 +1,37 @@
+# Principal Neighbourhood Aggregation for Graph Nets (PNA)
+
+[Principal Neighbourhood Aggregation for Graph Nets \(PNA\)](https://arxiv.org/abs/2004.05718)  is a graph learning model combining multiple aggregators with degree-scalers.
+
+
+### Datasets
+
+We perform graph classification experiment to reproduce paper results on [OGB](https://ogb.stanford.edu/). 
+
+### Dependencies
+
+- paddlepaddle >= 2.2.0
+- pgl >= 2.2.4
+
+### How to run
+
+
+```
+python main.py --config config.yaml   # train on ogbg-molhiv
+python main.py --config config_pcba.yaml # train on ogbg-molpcba
+```
+
+
+### Important Hyperparameters
+
+- aggregators: a list of aggregators name. ("mean", "sum", "max", "min", "var", "std")
+- scalers: a list of scalers name. ("identity", "amplification", "attenuation", "linear", "inverse_linear")
+- tower: The number of towers.
+- divide_input: hether the input features should be split between towers or not.
+- pre_layers: the number of MLP layers behind aggregators.
+- post_layers: MLP layers after aggregator.
+
+### Experiment results （ROC-AUC）
+|   | GIN   | PNA(paper result) | PNA(ours)|
+|-------------|----------|------------|-----------------|
+|HIV    | 0.7778  | 0.7905   | 0.7929     | 
+|PCBA   | 0.2266   | 0.2838   | 0.2801      |

examples/pna/config_pcba.yaml

@@ -40,6 +40,3 @@ log_filename: log.txt
 save_dir: ./checkpoints
 output_dir: ./outputs
 files2saved: ["*.yaml", "*.py", "./utils"]
-
-# python main_PCBA.py --type_net="complex" --batch_size=512 --lap_norm="none" --weight_decay=3e-6 --L=4 --hidden_dim=510 --out_dim=510 --residual=True 
-# --edge_feat=True  --readout=sum --graph_norm=True --batch_norm=True --aggregators="mean sum max" --scalers="identity" --config "configs/molecules_graph_classification_DGN_PCBA.json"  --lr_schedule_patience=4 --towers=5 --dropout=0.2 --init_lr=0.0005 --min_lr=0.00002 --edge_dim=16 --lr_reduce_factor=0.8

examples/pna/main.py

@@ -87,7 +87,7 @@ def main(config):
         num_workers=config.num_workers,
         collate_fn=fn)
 
-    deg_hog = get_degree_histogram(train_loader)
+    deg_hog = paddle.to_tensor(get_degree_histogram(train_loader))
 
     model = PNAModel(
         config.hidden_size,

examples/sag_pool/README.md

@@ -26,7 +26,7 @@ python main.py --use_cuda --dataset_name PROTEINS --lr 0.005 --batch_size 128 --
 
 - data\_path: the root path of your dataset 
 - dataset\_name: the name of the dataset. ("MUTAG", "IMDBBINARY", "IMDBMULTI", "COLLAB", "PROTEINS", "NCI1", "PTC", "REDDITBINARY", "REDDITMULTI5K")
-- fold\_idx: The $fold\_idx^{th}$ fold of dataset splited. Here we use 10 fold cross-validation
+- fold\_idx: The $fold\_{idx}^{th}$ fold of dataset splited. Here we use 10 fold cross-validation
 - min\_score: parameter for SAGPool which indicates minimal node score. (When min\_score is not None, pool\_ratio is ignored)
 - pool\_ratio: parameter for SAGPool which decides how many nodes will be removed.
 

pgl/nn/pna_conv.py

@@ -37,7 +37,7 @@ class PNAConv(nn.Layer):
         scalers: (list): List of scaler function keyword, 
             choices in ["identity", "amplification",
                         "attenuation", "linear", "inverse_linear"]
-        deg (numpy array): Histogram of in-degrees of nodes in the training set for computed avg_deg for scalers
+        deg (Tensor): Histogram of in-degrees of nodes in the training set for computed avg_deg for scalers
         towers (int, optional): Number of towers. Default: 1
         pre_layers (int, optional): Number of transformation layers before
             aggregation. Default: 1
@@ -77,12 +77,13 @@ def __init__(self,
 
         deg = deg.astype("float32")
         total_no_vertices = deg.sum()
-        bin_degrees = np.arange(len(deg)).astype("float32")
+        bin_degrees = paddle.arange(len(deg), dtype="float32")
         self.avg_deg = {
-            'lin': ((bin_degrees * deg).sum() / total_no_vertices),
-            'log': ((np.log(
-                (bin_degrees + 1)) * deg).sum() / total_no_vertices),
-            'exp': ((np.exp(bin_degrees) * deg).sum() / total_no_vertices),
+            'lin': ((bin_degrees * deg).sum() / total_no_vertices).item(),
+            'log':
+            (((bin_degrees + 1).log() * deg).sum() / total_no_vertices).item(),
+            'exp': (
+                (bin_degrees.exp() * deg).sum() / total_no_vertices).item(),
         }
         if use_edge:
             self.edge_mlp = paddle.nn.Linear(input_size, self.F_in)

tests/test_conv.py

@@ -109,7 +109,7 @@ def test_pna_conv(self):
                 "identity", "amplification", "attenuation", "linear",
                 "inverse_linear"
             ],
-            deg=np.asarray([0, 1, 1, 1, 2]),
+            deg=paddle.to_tensor([0, 1, 1, 1, 2]),
             towers=2,
             pre_layers=1,
             post_layers=2,