Score Conversion

Currently, the majority of outlier detectors are on node level. However, in some real-world applications, we may interest in the outlier edges or outlier graphs. In this tutorial, we introduce score converters provided in pygod.utils module, including to_edge_score and to_graph_score.

(Time estimate: 3 minutes)

Data Loading

We first load the data from PyGOD with load_data function.

from pygod.utils import load_data

data = load_data('weibo')
print(data)

Data(x=[8405, 400], edge_index=[2, 407963], y=[8405], train_mask=[8405], test_mask=[8405], val_mask=[8405])

Detector Training

Initialize and train a detector in PyGOD. Here, we use pygod.detector.DOMINANT as an example. For faster demonstration, we set epoch to 3.

from pygod.detector import DOMINANT

detector = DOMINANT(epoch=3)
detector.fit(data)

DOMINANT(act=<function relu at 0x7fa534639f80>,
         backbone=<class 'torch_geometric.nn.models.basic_gnn.GCN'>,
         batch_size=8405, compile_model=False, contamination=0.1,
         dropout=0.0, epoch=3, gpu=None, hid_dim=64, lr=0.004,
         num_layers=4, num_neigh=[-1, -1, -1, -1], save_emb=False,
         sigmoid_s=False, verbose=0, weight=0.5, weight_decay=0.0)

Obtaining Node Score

After training, we obtain raw outlier scores for each node with predict. The shape of node_score is (N, ).

node_score = detector.predict(data, return_pred=False, return_score=True)
print(node_score.shape)

torch.Size([8405])

Converting Score to Edge Level

To detect outlier edges, we convert the outlier scores on node level to edge level. The shape of edge_score is (E, ).

from pygod.utils import to_edge_score

edge_score = to_edge_score(node_score, data.edge_index)
print(edge_score.shape)

torch.Size([407963])

Converting Score to Graph Level

To detect outlier graphs, we convert the outlier scores on node level to graph level for each graph. graph_score is a scalar for each Data object. Here, we give an example for scoring a list of graph.

from pygod.utils import to_graph_score

data_list = [data, data, data]
graph_scores = []
for data in data_list:
    node_score = detector.predict(data, return_pred=False, return_score=True)
    graph_score = to_graph_score(node_score)
    graph_scores.append(graph_score.item())

print(graph_scores)

[19068.12890625, 19068.12890625, 19068.12890625]